Bioinformatics Tools For Pharmaceutical Drug Product Development 1St Edition Edition Vivek Chavda Full Chapter PDF

Bioinformatics Tools for
Pharmaceutical Drug Product

Development 1st Edition Edition Vivek
Chavda
Visit to download the full and correct content document:
https://ebookmass.com/product/bioinformatics-tools-for-pharmaceutical-drug-product-
development-1st-edition-edition-vivek-chavda/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Tools For Chemical Product Design From Consumer

Products to Biomedicine 1st Edition Mariano Martín
https://ebookmass.com/product/tools-for-chemical-product-design-
from-consumer-products-to-biomedicine-1st-edition-mariano-martin/
Chemoinformatics and bioinformatics in the

pharmaceutical sciences Pawan Kumar Raghav (Editor)
https://ebookmass.com/product/chemoinformatics-and-
bioinformatics-in-the-pharmaceutical-sciences-pawan-kumar-raghav-
editor/
Discovery and Development of Antidiabetic Agents from

Natural Products: Natural Product Drug Discovery Goutam
Brahmachari
https://ebookmass.com/product/discovery-and-development-of-
antidiabetic-agents-from-natural-products-natural-product-drug-
discovery-goutam-brahmachari/
Computation in BioInformatics S. Balamurugan
https://ebookmass.com/product/computation-in-bioinformatics-s-
balamurugan/
Process Systems Engineering for Pharmaceutical
Manufacturing 1st Edition Ravendra Singh
https://ebookmass.com/product/process-systems-engineering-for-
pharmaceutical-manufacturing-1st-edition-ravendra-singh/
Systems Architecture. Strategy and product development

for complex systems Bruce Cameron
https://ebookmass.com/product/systems-architecture-strategy-and-
product-development-for-complex-systems-bruce-cameron/
Practical Application of Supercritical Fluid

Chromatography for Pharmaceutical Research and
Development Hicks M.
https://ebookmass.com/product/practical-application-of-
supercritical-fluid-chromatography-for-pharmaceutical-research-
and-development-hicks-m/
Design of Hybrid Molecules for Drug Development Michael

Decker
https://ebookmass.com/product/design-of-hybrid-molecules-for-
drug-development-michael-decker/
Tools for Teaching 2nd Edition
https://ebookmass.com/product/tools-for-teaching-2nd-edition/
Bioinformatics Tools
for Pharmaceutical Drug
Product Development
Scrivener Publishing
100 Cummings Center, Suite 541J
Beverly, MA 01915-6106
Publishers at Scrivener
Martin Scrivener (martin@scrivenerpublishing.com)
Phillip Carmical (pcarmical@scrivenerpublishing.com)
Bioinformatics Tools
for Pharmaceutical Drug
Product Development
Edited by
Vivek Chavda
Department of Pharmaceutics and Pharmaceutical Technology,
L. M. College of Pharmacy, Ahmedabad, India
Krishnan Anand
Department of Chemical Pathology, School of Pathology, University of the Free
State, Bloemfontein, South Africa
and
Vasso Apostolopoulos
Institute for Health and Sport, Immunology and Translational Research Group,
Victoria University, Melbourne, Australia
This edition first published 2023 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA
© 2023 Scrivener Publishing LLC
For more information about Scrivener publications please visit www.scrivenerpublishing.com.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or other-
wise, except as permitted by law. Advice on how to obtain permission to reuse material from this title
is available at http://www.wiley.com/go/permissions.
Wiley Global Headquarters

111 River Street, Hoboken, NJ 07030, USA
For details of our global editorial offices, customer services, and more information about Wiley prod-
ucts visit us at www.wiley.com.
Limit of Liability/Disclaimer of Warranty

While the publisher and authors have used their best efforts in preparing this work, they make no rep
resentations or warranties with respect to the accuracy or completeness of the contents of this work and
specifically disclaim all warranties, including without limitation any implied warranties of merchant-
ability or fitness for a particular purpose. No warranty may be created or extended by sales representa
tives, written sales materials, or promotional statements for this work. The fact that an organization,
website, or product is referred to in this work as a citation and/or potential source of further informa
tion does not mean that the publisher and authors endorse the information or services the organiza
tion, website, or product may provide or recommendations it may make. This work is sold with the
understanding that the publisher is not engaged in rendering professional services. The advice and
strategies contained herein may not be suitable for your situation. You should consult with a specialist
where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other
commercial damages, including but not limited to special, incidental, consequential, or other damages.
Further, readers should be aware that websites listed in this work may have changed or disappeared
between when this work was written and when it is read.
Library of Congress Cataloging-in-Publication Data
ISBN 978-1-119-86511-7
Cover image: Pixabay.Com

Cover design by Russell Richardson
Set in size of 11pt and Minion Pro by Manila Typesetting Company, Makati, Philippines
Printed in the USA
10 9 8 7 6 5 4 3 2 1
Contents
Preface xv
Part I: Bioinformatics Tools 1

1 Introduction to Bioinformatics, AI, and ML
for Pharmaceuticals 3
Vivek P. Chavda, Disha Vihol, Aayushi Patel,
Elrashdy M. Redwan and Vladimir N. Uversky
1.1 Introduction 4
1.2 Bioinformatics 4
1.2.1 Limitations of Bioinformatics 8
1.2.2 Artificial Intelligence (AI) 8
1.3 Machine Learning (ML) 11
1.3.1 Applications of ML 12
1.3.2 Limitations of ML 14
1.4 Conclusion and Future Prospects 14
References 15
2 Artificial Intelligence and Machine Learning-Based
New Drug Discovery Process with Molecular Modelling 19
Isha Rani, Kavita Munjal, Rajeev K. Singla
and Rupesh K. Gautam
2.1 Introduction 20
2.2 Artificial Intelligence in Drug Discovery 21
2.2.1 Training Dataset Used in Medicinal Chemistry 22
2.2.2 Availability and Quality of Initial Data 23
2.3 AI in Virtual Screening 24
2.4 AI for De Novo Design 25
2.5 AI for Synthesis Planning 26
2.6 AI in Quality Control and Quality Assurance 27
2.7 AI-Based Advanced Applications 28
2.7.1 Micro/Nanorobot Targeted Drug Delivery System 28
v
vi Contents
2.7.2 AI in Nanomedicine 29
2.7.3 Role of AI in Market Prediction 29
2.8 Discussion and Future Perspectives 30
2.9 Conclusion 31
References 31
3 Role of Bioinformatics in Peptide-Based Drug Design
and Its Serum Stability 37
Vivek Chavda, Prashant Kshirsagar and Nildip Chauhan
3.1 Introduction 37
3.2 Points to be Considered for Peptide-Based Delivery 38
3.3 Overview of Peptide-Based Drug Delivery System 40
3.4 Tools for Screening of Peptide Drug Candidate 41
3.5 Various Strategies to Increase Serum Stability of Peptide 42
3.5.1 Cyclization of Peptide 42
3.5.2 Incorporation of D Form of Amino Acid 44
3.5.3 Terminal Modification 44
3.5.4 Substitution of Amino Acid Which is Not Natural 46
3.5.5 Stapled Peptides 46
3.5.6 Synthesis of Stapled Peptides 47
3.6 Method/Tools for Serum Stability Evaluation 47
3.7 Conclusion 48
3.8 Future Prospects 49
References 49
4 Data Analytics and Data Visualization
for the Pharmaceutical Industry 55
Shalin Parikh, Ravi Patel, Dignesh Khunt, Vivek P. Chavda
and Lalitkumar Vora
4.1 Introduction 56
4.2 Data Analytics 57
4.3 Data Visualization 58
4.4 Data Analytics and Data Visualization for Formulation
Development 60
4.5 Data Analytics and Data Visualization for Drug
Product Development 65
4.6 Data Analytics and Data Visualization for Drug
Product Life Cycle Management 69
References 72
Contents vii
5 Mass Spectrometry, Protein Interaction and Amalgamation

of Bioinformatics 77
Vivek Chavda, Kaustubh Dange and Madhav Joglekar
5.1 Introduction 77
5.2 Mass Spectrometry - Protein Interaction 79
5.2.1 The Prerequisites 80
5.2.2 Finding Affinity Partner (The Bait) 80
5.2.3 Antibody-Based Affinity Tags 80
5.2.4 Small Molecule Ligands 80
5.2.5 Fusion Protein-Based Affinity Tags 81
5.3 MS Analysis 81
5.4 Validating Specific Interactions 82
5.5 Mass Spectrometry – Qualitative and Quantitative Analysis 83
5.6 Challenges Associated with Mass Analysis 83
5.7 Relative vs. Absolute Quantification 85
5.8 Mass Spectrometry – Lipidomics
and Metabolomics 86
5.9 Mass Spectrometry – Drug Discovery 87
5.10 Conclusion and Future Scope 88
5.11 Resources and Software 89
Acknowledgement 89
References 89
6 Applications of Bioinformatics Tools in Medicinal
Biology and Biotechnology 95
Harshil Shah, Vivek Chavda and Moinuddin M. Soniwala
6.1 Introduction 96
6.2 Bioinformatics Tools 97
6.3 The Genetic Basis of Diseases 97
6.4 Proteomics 98
6.5 Transcriptomic 100
6.6 Cancer 101
6.7 Diagnosis 102
6.8 Drug Discovery and Testing 103
6.9 Molecular Medicines 105
6.10 Personalized (Precision) Medicines 106
6.11 Vaccine Development and Drug Discovery
in Infectious Diseases and COVID-19 Pandemic 108
6.12 Prognosis of Ailments 109
6.13 Concluding Remarks and Future Prospects 110
viii Contents
Acknowledgement 111
References 111
7 Clinical Applications of “Omics” Technology
as a Bioinformatic Tool 117
Vivek Chavda, Rajashri Bezbaruah, Disha Valu, Sanjay Desai,
Nildip Chauhan, Swati Marwadi, Gitima Deka
and Zhiyong Ding
Abbreviations 118
7.1 Introduction 118
7.2 Execution Method 119
7.3 Overview of Omics Technology 121
7.4 Genomics 124
7.5 Nutrigenomics 127
7.6 Transcriptomics 128
7.7 Proteomics 129
7.8 Metabolomics 130
7.9 Lipomics or Lipidomics 133
7.10 Ayurgenomics 134
7.11 Pharmacogenomics 134
7.12 Toxicogenomic 135
Acknowledgement 139
References 139
Part II: Bioinformatics Tools for Pharmaceutical

Sector 147
8 Bioinformatics and Cheminformatics Tools
in Early Drug Discovery 149
Palak K. Parikh, Jignasa K. Savjani, Anuradha K. Gajjar
and Mahesh T. Chhabria
Abbreviations 150
8.2 Informatics and Drug Discovery 152
8.3 Computational Methods in Drug Discovery 153
8.3.1 Homology Modeling 153
8.3.2 Docking Studies 155
8.3.3 Molecular Dynamics Simulations 158
8.3.4 De Novo Drug Design 159
8.3.5 Quantitative Structure Activity Relationships 160
8.3.6 Pharmacophore Modeling 161
Contents ix
8.3.7 Absorption, Distribution, Metabolism, Excretion

and Toxicity Profiling 165
8.4 Conclusion 168
References 169
Formulation and Process Development for Drug Products 183
Vivek P. Chavda
9.2 Current Scenario in Pharma Industry and Quality
by Design (QbD) 185
9.3 AI- and ML-Based Formulation Development 187
9.4 AI- and ML-Based Process Development
and Process Characterization 189
9.5 Concluding Remarks and Future Prospects 192
References 193
Manufacturing and Drug Product Marketing 197
Kajal Baviskar, Anjali Bedse, Shilpa Raut
and Narayana Darapaneni
Abbreviations 198
10.1 Introduction to Artificial Intelligence
and Machine Learning 199
10.1.1 AI and ML in Pharmaceutical Manufacturing 200
10.1.2 AI and ML in Drug Product Marketing 201
10.2 Different Applications of AI and ML in the Pharma Field 202
10.2.1 Drug Discovery 202
10.2.2 Pharmaceutical Product Development 202
10.2.3 Clinical Trial Design 203
10.2.4 Manufacturing of Drugs 203
10.2.5 Quality Control and Quality Assurance 203
10.2.6 Product Management 203
10.2.7 Drug Prescription 204
10.2.8 Medical Diagnosis 204
10.2.9 Monitoring of Patients 204
10.2.10 Drug Synergism and Antagonism Prediction 204
10.2.11 Precision Medicine 205
10.3 AI and ML-Based Manufacturing 205
10.3.1 Continuous Manufacturing 205
10.3.2 Process Improvement and Fault Detection 209
x Contents
10.3.3 Predictive Maintenance (PdM) 210

10.3.4 Quality Control and Yield 211
10.3.5 Troubleshooting 211
10.3.6 Supply Chain Management 212
10.3.7 Warehouse Management 213
10.3.8 Predicting Remaining Useful Life 214
10.3.9 Challenges 215
10.4 AI and ML-Based Drug Product Marketing 217
10.4.1 Product Launch 217
10.4.2 Real-Time Personalization and Consumer
Behavior 218
10.4.3 Better Customer Relationships 219
10.4.4 Enhanced Marketing Measurement 220
10.4.5 Predictive Marketing Analytics 220
10.4.6 Price Dynamics 221
10.4.7 Market Segmentation 222
10.4.8 Challenges 223
10.5 Future Prospects and Way Forward 223
10.6 Conclusion 224
References 225
11 Artificial Intelligence and Machine Learning Applications
in Vaccine Development 233
Ali Sarmadi, Majid Hassanzadeganroudsari and M. Soltani
11.2 Prioritizing Proteins as Vaccine Candidates 237
11.3 Predicting Binding Scores of Candidate Proteins 238
11.4 Predicting Potential Epitopes 243
11.5 Design of Multi-Epitope Vaccine 244
11.6 Tracking the RNA Mutations of a Virus 245
Conclusion 248
References 249
12 AI, ML and Other Bioinformatics Tools for Preclinical
and Clinical Development of Drug Products 255
Avinash Khadela, Sagar Popat, Jinal Ajabiya, Disha Valu,
Shrinivas Savale and Vivek P. Chavda
Abbreviations 256
12.2 AI and ML for Pandemic 258
12.3 Advanced Analytical Tools Used in Preclinical
and Clinical Development 259
Contents xi
12.3.1 Spectroscopic Techniques 260

12.3.2 Chromatographic Techniques 263
12.3.3 Electrochemical Techniques 263
12.3.4 Electrophoretic Techniques 264
12.3.5 Hyphenated Techniques 264
12.4 AI, ML, and Other Bioinformatics Tools for Preclinical
Development of Drug Products 265
12.4.1 Various Computational Tools Used
in Pre-Clinical Drug Development 266
12.5 AI, ML, and Other Bioinformatics Tools for Clinical
Development of Drug Products 268
12.5.1 Role of AI, ML, and Bioinformatics in Clinical
Research 270
12.5.2 Role of AI and ML in Clinical Study Protocol
Optimization 272
12.5.3 Role of AI and ML in the Management
of Clinical Trial Participants 272
12.5.4 Role of AI and ML in Clinical Trial Data
Collection and Management 272
12.6 Way Forward 275
12.7 Conclusion 276
References 277
Part III: Bioinformatics Tools for Healthcare

Sector 285
13 Artificial Intelligence and Machine Learning
in Healthcare Sector 287
Vivek P. Chavda, Kaushika Patel, Sachin Patel
and Vasso Apostolopoulos
Abbreviations 288
13.2 The Exponential Rise of AI/ML Solutions in Healthcare 289
13.3 AI/ML Healthcare Solutions for Doctors 291
13.4 AI/ML Solution for Patients 293
13.5 AI Solutions for Administrators 295
13.6 Factors Affecting the AI/ML Implementation
in the Healthcare Sector 297
13.6.1 High Cost 297
13.6.2 Lack of Creativity 298
13.6.3 Errors Potentially Harming Patients 298
13.6.4 Privacy Issues 298
xii Contents
13.6.5 Increase in Unemployment 299

13.6.6 Lack of Ethics 299
13.6.7 Promotes a Less-Effort Culture Among Human
Workers 299
13.7 AI/ML Based Healthcare Start-Ups 299
13.8 Opportunities and Risks for Future 304
13.8.1 Patient Mobility Monitoring 305
13.8.2 Clinical Trials for Drug Development 305
13.8.3 Quality of Electronic Health Records (EHR) 305
13.8.4 Robot-Assisted Surgery 305
13.9 Conclusion and Perspectives 306
References 307
14 Role of Artificial Intelligence in Machine Learning
for Diagnosis and Radiotherapy 315
Sanket Chintawar, Vaishnavi Gattani, Shivanee Vyas
and Shilpa Dawre
Abbreviations 316
14.2 Machine Learning Algorithm Models 318
14.2.1 Supervised Learning 318
14.2.2 Unsupervised Learning 319
14.2.3 Semi-Supervised Learning 319
14.2.4 Reinforcement Learning (RL) 320
14.3 Artificial Learning in Radiology 321
14.3.1 Types of Radiation Therapy 321
14.3.1.1 External Radiation Therapy 322
14.3.1.2 Internal Radiation Therapy 323
14.3.1.3 Systemic Radiation Therapy 323
14.3.2 Mechanism of Action 323
14.4 Application of Artificial Intelligence and Machine
Learning in Radiotherapy 324
14.4.1 Delineation of the Target 324
14.4.2 Radiotherapy Delivery 325
14.4.3 Image Guided Radiotherapy 327
14.5 Implementation of Machine Learning Algorithms
in Radiotherapy 328
14.5.1 Image Segmentation 328
14.5.2 Medical Image Registration 329
14.5.3 Computer-Aided Detection (CAD)
and Diagnosis System 329
14.6 Deep Learning Models 331
Contents xiii
14.6.1 Deep Neural Networks 331

14.6.2 Convolutional Neural Networks 332
14.7 Clinical Implementation of AI in Radiotherapy 332
14.8 Current Challenges and Future Directions 339
References 339
15 Role of AI and ML in Epidemics and Pandemics 345
Rajashri Bezbaruah, Mainak Ghosh, Shuby Kumari,
Lawandashisha Nongrang, Sheikh Rezzak Ali, Monali Lahiri,
Hasmi Waris and Bibhuti Bhushan Kakoti
15.2 History of Artificial Intelligence (AI) in Medicine 347
15.3 AI and MI Usage in Pandemic and Epidemic (COVID-19) 348
15.3.1 SARS-CoV-2 Detection and Therapy Using
Machine Learning and Artificial Intelligence 349
15.3.2 SARS-Cov-2 Contact Tracing Using Machine
Learning and Artificial Intelligence 350
15.3.3 SARS-CoV-2 Prediction and Forecasting Using
15.3.4 SARS-CoV-2 Medicines and Vaccine Using
15.4 Cost Optimization for Research and Development
Using Al and ML 351
15.5 AI and ML in COVID 19 Vaccine Development 352
15.6 Efficacy of AI and ML in Vaccine Development 357
15.7 Artificial Intelligence and Machine Learning
in Vaccine Development: Clinical Trials During
an Epidemic and Pandemic 358
15.8 Clinical Trials During an Epidemic 360
15.8.1 Ebola Virus 360
15.8.2 SARS-CoV-2 361
15.9 Conclusion 361
References 362
16 AI and ML for Development of Cell and Gene Therapy
for Personalized Treatment 371
Susmit Mhatre, Somanshi Shukla, Vivek P. Chavda,
Lakshmikanth Gandikota and Vandana Patravale
16.1 Fundamentals of Cell Therapy 372
16.1.1 Stem Cell Therapies 374
16.1.1.1 Mesenchymal Stem Cells (MSCs) 375
16.1.1.2 Hematopoietic Stem Cells (HSCs) 375
xiv Contents
16.1.1.3 Mononuclear Cells (MNCs) 375

16.1.1.4 Endothelial Progenitor Cells (EPCs) 375
16.1.1.5 Neural Stem Cells (NSCs) or Neural
Progenitor Cells (NPCs) 376
16.1.2 Adoptive Cell Therapy 376
16.1.2.1 Tumor-Infiltrating Lymphocyte (TIL)
Therapy 376
16.1.2.2 Engineered T-Cell Receptor (TCR)
Therapy 377
16.1.2.3 Chimeric Antigen Receptor (CAR)
T Cell Therapy 377
16.1.2.4 Natural Killer (NK) Cell Therapy 377
16.2 Fundamentals of Gene Therapy 378
16.2.1 Identification 378
16.2.2 Treatment 379
16.3 Personalized Cell Therapy 381
16.4 Manufacturing of Cell and Gene-Based Therapies 382
16.5 Development of an Omics Profile 385
16.6 ML in Stem Cell Identification, Differentiation,
and Characterization 387
16.7 Machine Learning in Gene Expression Imaging 389
16.8 AI in Gene Therapy Target and Potency Prediction 390
16.9 Conclusion and Future Prospective 391
References 392
17 Future Prospects and Challenges in the Implementation
of AI and ML in Pharma Sector 401
Prashant Pokhriyal, Vivek P. Chavda and Mili Pathak
17.1 Current Scenario 402
17.2 Way Forward 406
References 407
Index 417
Preface
For a new drug to be developed and brought to market, approximately

US$1.8 billion and a minimum of 15 years in development are required. In
most instances, only a few drugs make it to market because the process of
creating a new drug can fail at different steps along the way, with most of
them failing in the final stages of development. Some reasons for this can
be attributed to the lack of extensive clinical data, unexpected toxicities
and long-term side effects; as well as the highly competitive market, which
puts a strain on the development of new drugs. A way to reduce some of the
costs and increase the likelihood of success is to maximize the information
gained via basic science and the design of better translational approaches
and clinical trials. As such, bioinformatics approaches are becoming more
essential in drug discovery and vaccine design, not only in academia,
but also in the pharmaceutical industry. Bioinformatics involves the use
of software tools and computer programming to understand biological
data, particularly when the data is large and complex. The development
of large data warehouses and algorithms to analyze large data, the iden-
tification of biomarkers and novel drug targets, computational biochem-
istry, genomics, drug discovery and design have all been at the forefront
of translational drug discovery in recent years. Bioinformatics has revo-
lutionized disease-based and drug-based approaches as well as improved
knowledge of biological targets. It has ushered in a new era of research with
the aim to accelerate the design and development of drug and vaccine tar-
gets, improve validation approaches as well as facilitate in identifying side
effects and predict drug resistance. As such, this will aid in more successful
drug candidates from discovery to clinical trials to the market and, most
importantly, make it a more cost-effective process overall.
Since there has been much emphasis placed on developing bioinformat-
ics tools for pharmaceutical drug development, this book is a timely and
important addition to the evolving field. The 17 chapters are categorized
into three sections. The first section presents the latest information on bio-
informatics tools, artificial intelligence, machine learning, computational
xv
xvi Preface
methods, protein interactions, peptide-based drug design and omics tech-

nologies. The following two sections include bioinformatics tools for the
pharmaceutical and healthcare sectors.
In this book, experts from around the world provide comprehen-
sive overviews of the many important steps involved in—and the critical
insights needed for—the successful development of therapeutic drug prod-
ucts with the help of bioinformatics, artificial intelligence and machine
learning. The amount of work put into these 17 chapters required signifi-
cant collaboration and input, and there are many people who are worthy
of our thanks. We thank the support of over 20 thought leaders in AI/
ML from across the globe who contributed to the chapters. Without their
contributions, the book would not have been possible. We offer our sin-
cere gratitude to the Department of Pharmaceutics and Pharmaceutical
Technology, L.M. College of Pharmacy, Ahmedabad-Gujarat, India; the
Department of Chemical Pathology, School of Pathology, University of the
Free State, Bloemfontein campus, Free State, South Africa; and Victoria
University, Institute for Health and Sport for their ongoing support in pub-
lishing this volume. We also thank Scrivener Publishing and their staff for
their editorial support throughout the publication process.
The Editors
Vivek Chavda
K. Anand
Vasso Apostolopoulos
December 2022
Part I
BIOINFORMATICS TOOLS
1
Introduction to Bioinformatics, AI,
and ML for Pharmaceuticals
Vivek P. Chavda1*, Disha Vihol2, Aayushi Patel3, Elrashdy M. Redwan4,5
and Vladimir N. Uversky6†
1
Department of Pharmaceutics and Pharmaceutical Technology, L. M. College of
Pharmacy, Ahmedabad, Gujarat, India
2
Department of Phytopharmacy and Phytomedicine, School of Pharmacy, Gujarat
Technological University, Ahmedabad, Gujarat, India
3
Pharmacy Section, L. M. College of Pharmacy, Ahmedabad, Gujarat, India
4
Department of Biological Sciences, Faculty of Sciences, King Abdulaziz University,
Jeddah, Saudi Arabia
5
Therapeutic and Protective Proteins Laboratory, Protein Research Department,
Genetic Engineering and Biotechnology Research Institute, City of Scientific
Research and Technological Applications, New Borg EL-Arab,
Alexandria, Egypt
6
Department of Molecular Medicine and Byrd Alzheimer’s Research Institute,
Morsani College of Medicine, University of South Florida, Tampa, FL, USA
Abstract
Bioinformatics is a growing field that has emerged in recent years. The use of com-
putational applications for protein sequence analysis in the early 1960s created the
groundwork for bioinformatics. Alongside, developments in molecular biology
techs evolved DNA analysis, leading to simpler manipulation of DNA, its sequenc-
ing, and computer science, suggesting the development of compatible and more
powerful computers with innovative software for performing bioinformatics tasks.
Biological Big Data collection when analyzed with bioinformatics tools leads to
powerful predictive results with repeatability. Due to advancements in the merg-
ing of computer science and biology, even subdisciplines like synthetic biology,
systems biology, and whole-cell modeling are emerging rapidly.
*Corresponding author: vivek7chavda@gmail.com

†
Corresponding author: vuversky@usf.edu: ORCID: https://orcid.org/0000-0002-4037-5857
Vivek Chavda, Krishnan Anand and Vasso Apostolopoulos (eds.) Bioinformatics Tools for Pharmaceutical
Drug Product Development, (3–18) © 2023 Scrivener Publishing LLC
3
4 Bioinformatics Tools in Drug Development
Keywords: Bioinformatics, artificial intelligence, machine learning, AI/ML,

pharmaceuticals, drug product development
1.1 Introduction
In the context of Artificial Intelligence (AI), Machine Learning (ML), and
Big Data, the healthcare industry explores the medication research process,
evaluating how emerging technologies can enhance efficacy [1]. Artificial
intelligence and machine learning are seen as the way of the future in a
variety of fields and sectors, including pharmaceuticals. In a world, where
a single authorized medicine costs millions of dollars and requires years of
rigorous testing before being licensed, saving money and time is a priority.
Producing new pharmacological compounds to combat any disease is
an expensive and time-consuming procedure, yet it goes unchecked. The
most important aspect of drug design is to take advantage of the collected
data and seek fresh and unique leads. Once the medication target has been
determined, numerous multidisciplinary domains collaborate to develop
enhanced pharmaceuticals using AI and ML technologies [2]. These tech-
nologies are utilized at every phase of the computer-aided drug discovery
process, and combining them results in a proven track record of success in
finding hit molecules. Furthermore, the fusion of AI and ML with high-
dimension data enhanced the capabilities of computer-aided drug discov-
ery and design [3]. Clinical trial output prediction using AI/ML integrated
models might decrease the costs of the clinical trial, while simultaneously
increasing their success rates. In this study, we will be covering the poten-
tial of AI and ML technologies in enabling computer-aided drug creation,
along with challenges and opportunities for the pharmaceutical sector.
1.2 Bioinformatics
When biological data along with genetic information is analyzed using
computer technology for calculating and obtaining mathematically and
statistically approved results, is called Bioinformatics. The computational
means are utilized for addressing data-intensive, large-scale biological
challenges [4]. It includes the development and application of databases,
algorithms, computational and statistical tools, and theory to tackle for-
mal and practical issues emerging from biological data administration and
analysis [5, 6].
Bioinformatic Tools for Pharmaceuticals 5
Bioinformatics allows rapid molecular modeling of biological processes

from the collected big data for obtaining meaningful conclusions through
various stages such as compilation of the statistical information from bio-
logical data, creation of a computational model, the resolution of computa-
tional modeling issues, and the assessment of the resulting computational
algorithm [7]. Genomics and proteomics are the branches of bioinformat-
ics that aim at understanding the organizing principles encoded inside the
nucleic acid and protein sequences, respectively. Image and signal process-
ing enables usable conclusions to be extracted from vast volumes of raw
data [5]. It helps in decoding genetics by facilitating text mining of bio-
logical literature, comparing, analyzing, and interpreting genetic, genomic,
and proteomic data, assessing gene and protein expression, detecting
mutations, sequencing and annotating genomes, and interpreting evolu-
tionary aspects with disease pathways [8, 9]. Moreover, it helps to simulate
and model DNA, RNA, proteins, and biomolecular interactions in struc-
tural biology. All of these can be achieved by correlating the biological data
for understanding the effect of the diseased condition on the body’s nor-
mal cellular functions [10]. Hence, bioinformatics has progressed now to
the point, where analysis and interpretation of diverse forms of data is the
most important challenge.
Omics technologies offer a chance to investigate changes in DNA, RNA,
proteins, and other biological components across intraspecies and inter-
species. The analysis of these components is interesting from a toxicity
assessment view as they can alter in response to chemical or drug exposure
in cells or tissues. Genomic research, which generates enormous amounts
of data, is one area, where bioinformatics is very valuable. Bioinformatics
aids in the interpretation of data, which may then be used to provide a
diagnosis for a patient suffering from a rare ailment, track and monitor
infectious organisms as they spread across a community, or determine the
best therapy for a cancer patient [11]. Genomics sequences assemble and
analyze the structure and function of genomes using recombinant DNA,
DNA sequencing technologies, and bioinformatics. Various software used
in bioinformatics includes accessibility of protein surface and secondary
structure predictions using NetSurfP, prediction of beta-turn sites in pro-
tein sequences using NetTurnP, and AutoDock for Automated Docking
Tool Suite [4].
Rapid advances in genomics and other molecular research tools, along
with advances in information technology, have resulted in a massive vol-
ume of molecular biology knowledge during the last few decades [12].
Bioinformatics will continue to advance as a result of the integration of
many technologies and techniques that bring together professionals from
Table 1.1 Various bioinformatics and AI-driven tools applied in the pharmacy
department and industry.
Computational
tools Application Reference
BLAST The Basic Local Alignment Search Tool [15]
(BLAST) is used for searching local
similarity regions between sequences and
compares to the available database for
calculating the statistical significance of
matches. The matching infers functional
and evolutionary relationships between
sequences and identifies genetically
related families.
ChEMBL ChEMBL is designed manually to maintain [16]
a database of bioactive molecules. It
correlates genomic data with chemical
structure and bioactivity for the
development of new drugs.
geWorkbench The genomics Workbench (geWorkbench) [11]
comprises the tools for performing
management, analysis, visualization, and
annotation of biomedical data. It supports
data for microarray gene expression,
DNA and Protein Sequences, pathways,
Molecular structure – prediction,
Sequence Patterns, Gene Ontology, and
Regulatory Networks
GROMACS It is software for high-performance molecular [7]
dynamics and output analysis, especially for
proteins and lipids.
IGV Integrative Genomics Viewer (IGV) is [17]
a high-performance, user-friendly,
interactive tool for the visualization and
exploration of genomic data.
MODELLER A protein three-dimensional structural [18]
homology or comparative modeling tool.
SwissDrugDesign SwissDrugDesign provides a collection of [19]
tools required for Computer-Aided Drug
Design (CADD).
(Continued)
Table 1.1 Various bioinformatics and AI-driven tools applied in the pharmacy
department and industry. (Continued)
Computational
tools Application Reference
UCSF Chimera UCSF Chimera is an interactive tool for the [20]
visualization and analysis of molecular
structures.
AlphaFold It is an AI system developed to [21]
computationally predict protein
structures with unprecedented accuracy
and speed.
Cyclica The ML tool address challenges faced [21]
across the drug discovery life cycle
by correlating biophysics, medicinal
chemistry, and systems biology.
DeepChem It is a deep learning framework for drug [18]
discovery.
DeltaVina Gives docking scores for protein-ligand [17]
binding affinity
Exscientia The AI engineers precision medicines more [13]
rapidly and efficiently by accelerating
pre-clinical drug discovery phases
through monitoring and analysis of drug
design and experiments.
Hit Dexter It estimates how likely a small molecule [4]
is to trigger a positive response in
biochemical and biological assays.
ORGANIC Objective-Reinforced Generative [8]
Adversarial Network for Inverse-
design Chemistry (ORGANIC) is a
tool for creating molecules with desired
properties.
Somatix It is a real-time gesture detection [22]
technology that enables passive data
collection of indoor and outdoor
patients for enhancing medication
adherence rates, data reliability, and
cost-effectiveness.
other domains to build cutting-edge computational and informational

tools tailored to the biomedical research business [4, 9]. Table 1.1. rep-
resents various bioinformatics and AI-driven tools, which can be applied
in the pharmacy department and industry.
1.2.1 Limitations of Bioinformatics

In the case of drug discovery by applying in silico approaches, prediction
of ligand affinity and inhibitory activity, where adequate training data are
available is possible at certain levels but search for the mechanism-based
inhibitors remains a highly challenging task. In structure-based approaches
to various binding affinities, free energy calculations without extensive
data collection at times leads to adequate results but high chances of false-
positive rates remain a limiting factor. While chemical structure mining,
the bioinformatics tools are efficiently able to produce a large number of
metabolites but it is by no means possible to find ways of ranking metabo-
lites accurately. Prediction of metabolic rates is generally not possible with
bioinformatics tools. The in vivo predictions by applying bioinformatics for
assessing biological activity and toxicological effects detect most toxico-
phores but the prediction of time-dependent inhibitors remains a difficult
task [13].
Furthermore, it requires a sophisticated laboratory for the in-depth study
and collection of biomolecules, and the establishment of such laboratories
requires significant funds. The system operation is a complex mechanism
so specially trained individuals are required for handling computer-based
biological data. Uninterrupted electricity supply for biological investiga-
tions using computational applications is the basic requirement, abrupt
interruption can lose huge data from the system memory. The system must
also be virus-protected, which, if not controlled, can lead to the deletion of
data and corruption of the programs [14].
1.2.2 Artificial Intelligence (AI)

Artificial intelligence (AI) is a branch of computer science that unlike nat-
ural intelligence is demonstrated by machines or computers and is defined
as a system that analyzes its environment and performs a series of actions
to achieve goals similar to a human-being process. The objectives of AI
include gathering information, establishing rules on usage of the infor-
mation, reasoning, data representation, organizing, planning, learning,
problem-solving, processing, and the ability to manipulate data or self-

correction [9, 21].
In the pharmaceutical industry, AI plays role in providing technologies
for drug discovery, such as drug target identification, optimization, design-
ing, formulation development, manufacturing, break-down R&D costs,
analyzing biomedicine information for recruiting suitable patients for clin-
ical trials, drug repurposing, diagnostic tools and assistance, and optimiz-
ing medical treatment processes [23]. Moreover, AI optimizes innovation,
enhances the efficiency of research by data management, and creates com-
putational tools for researchers, physicians, and regulators with minimiz-
ing human intervention and errors [24]. Generally, two main classes of AI
technology developments are incorporated in the case of drug discovery.
The first component comprises computing methodologies including sys-
tems simulating human experiences along with outputs. The second one
consists of models mimicking Artificial Neural Networks (ANNs) for real-
time data correlation with the management of AI technology evolution
[25].
Various in silico models predict pharmacokinetics and simulate
molecular docking of the drugs to ease down drug discovery phases
along with predictions of in vitro and in vivo responses [26, 27]. AI in
drug development includes predictions of probable synthetic pathways
for drug-like molecules, pharmacokinetic and dynamic properties, pro-
tein identification and characterization, bioactivity, drug-target, and
drug-drug interactions [18, 28, 29]. Moreover, AI incorporates various
omics for identifying new disease pathways with targets using novel
biomarkers and therapeutic targets [30]. In the case of clinical trials,
AI improves candidate selection criteria by identifying the best patients
with human-relevant biomarkers of disease and gene targets for the study
and ensuring the most suitable trial results. It also helps in removing
the trail hindering elements and reduces the time for conducting large
database analysis [24]. An example of such an AI platform is AiCure,
a mobile application subjected to phase 2 clinical trial patients with
schizophrenia for assessing improvement in patient medication adher-
ence [31]. The AI-driven PAT (Process analytical technology) proves
to be a necessary tool in terms of quality control and assurance while
manufacturing. Improvements can be observed in product yield, utiliza-
tion, and cost-saving with less waste generation [32]. To assess real-time
manufacturing aspects, the Manufacturing Execution System (MES) is
utilized. It complies with regulatory guidelines and ensures high-quality
product development through risk management, shortened production

cycles, optimized resource use, and controlled batch release [33]. Then
the Automatic Process Control System (APCS) is used for ensuring a safe
and profitable process by monitoring and optimizing process variables
like concentration, flow, pressure, temperature, and vacuum [34].
Furthermore, AI can be utilized in drug repurposing, where a drug gets
qualified to enter Phase II trials for different use without going through
Phase I clinical trials and toxicology testing which reduces costing and
time of the trials [35]. Not only AI can be used for drug discovery, but
it can also be used in polypharmacology, the ‘one-disease–multiple tar-
gets theory. Databases such as BindingDB, ChEMBL, DrugBank, Ligand
Expo, PubChem, PDB, and ZINC are available for information on binding
affinities, biological activities, chemical properties, crystal structures, drug
targets, and pathways [36].
The digitalization of health and medicine has created an opportunity for
AI in hospital pharmacies for performing tasks, such as maintaining elec-
tronic medical records and patients’ medication history, designing treat-
ment approaches and dosage forms, medication safety, drug interactions,
ADME consultations, and providing healthcare aid to patients. This way AI
agrees with share-risk agreements and decision-making in Pharmacy and
Therapeutics Committees [37, 38]. Electronic health records (EHR) are
collected routinely and can be classified into structured and unstructured
data. Structured data refers to the collection of data in an organized man-
ner with standard units and ranges (e.g., vital signs), unlike unstructured
data with unclear management (e.g., imaging results) [39]. This data is col-
lected by AI in real-time for analyzing clinical data management and prac-
tice which can give insights into novel drug discovery, pharmacovigilance,
drug-associated adverse events, patient medication adherence, and pre-
scription errors [40]. With the information of EHR, various patient-omics
data (i.e., genomics, microbiomics, proteomics) can be integrated for the
creation of the Electronic Medical Records and Genomics (eMERGE) net-
work which helps in identifying unknown diseases with associations to the
gene bank obtained [41]. AI can also predict an epidemic outbreak. Even
predictions of shipment times of therapeutics can be carried out efficiently
by incorporating AI tools in the case of e-pharmacy. Moreover, AI can be
used as a diagnostic tool for disease analysis and status by grading system
with reproducibility. It improves the accuracy of the treatment decisions
and predicts prognosis. Even data can be collected from uncooperative
patients [42].
Advantages of AI include providing real-time data management, error

minimization and producing efficient output, multitasking, patient data
management, adverse effects or side effects data collection, medication
designing with disease correlation, streamlining tasks, inventory manage-
ment, and assisting research in the development of drug delivery formula-
tions. Nevertheless, disadvantages are also a part of the AI and they are the
need for human surveillance, expensive building and launching of AI tools,
chances of false report generation, lack of data collection and method stan-
dardization, may overlook social variables, raises unemployment, no cre-
ativity, the risk of data leakage and mass-scale destruction, and acceptance
within the healthcare sector [37, 43].
1.3 Machine Learning (ML)

Machine learning (ML) is a subclass of AI, where algorithms process big
data to detect patterns, learn from them, and solve problems statistically
and autonomously. ML is categorized into supervised, unsupervised, and
reinforcement learning. Supervised learning includes the application of
regression and classification methods which forms a predictive model
upon data insertion from input and output sources. The predictive models
can be disease diagnosis or drug efficacy and ADMET predictions [44]. In
unsupervised mode, solely input data are utilized and interpreted using
clustering and feature-finding methods. The output comprises discovering
a disease with its probable targets [45, 46]. Lastly, reinforcement learning
depends majorly on decision-making in the applied or specific environ-
ment with maximum performance ability. By applying modeling and
quantum chemistry, outputs such as de novo drug design can be executed
with this learning mode [47].
ML includes a subdivision consisting of Deep Learning (DL), which
engages Artificial Neural Networks (ANNs) for adapting and learning the
experimental data. These networks are similar to human biological neu-
rons responsible for electrical impulses transmission in the brain, which
allows real-time data collection and interpretation [48]. This big data in
association with algorithm application can help in discovering new drugs
with more potency and can improve the personalized medicine sector
based on genetic markers [47].
Machine learning in healthcare performs multiple tasks, such as
classification, recommendations, clustering and correlation of cases,
prediction, anomaly detection, automation, and ranking of information

[49]. The disease progression and development mechanism within a
body is a complex system that cannot be understood by simple data col-
lection. Usually, real-time data collection and compilation are carried out
by high throughput approaches such as the usage of the pre-defined set
of machine learning applications. This software not only provides diag-
nostic approaches but also helps in identifying hypothetical therapeutics
for drug development. The benefits of incorporating machine learning
are infinite availability of data storage and high flexibility in its manage-
ment. Various data sets include assay information, biometrics, images,
omics data, textual information, and data collected from wearables [50,
51]. Various ML applications such as Python, Spark, MLLib, and Jupyter
notebooks have been utilized by pharma industries for data mining and
predictive intelligence for solving daily tasks along with moderately
tedious challenges [52].
1.3.1 Applications of ML
a) Research and development of new drug:
ML utilizes a feedback-driven drug development process by interpreting
existing results from sources, such as computational modeling data, liter-
ature surveys, and high-throughput screening. This process helps in iden-
tifying lead compounds with efficiency and reduced randomness, errors,
and time-lapse. In the approach such as de novo design, inputs require a
compound library gained through in silico methods and virtual screening
applications which mimic bioactivity and toxicity models [53]. The drug
discovery can be carried out by following a series of steps starting from
the identification of novel bioactive compounds through docking studies
and molecular dynamics. A hit compound can be found while screening
chemical libraries, computer simulation, or screening naturally isolated
materials, such as plants, bacteria, and fungi. Then the recognized hits
are screened in cell-based assays consisting of animal models of disease
to assess efficacy and safety. Once the activity of the lead molecule is con-
firmed, chemical modifications can be carried out in search of a novel com-
pound consisting of maximal therapeutic benefits with minimal harm [54,
55]. Hence, incorporating algorithm datasets in conjugation with chemical
structures and targets is utilized for the optimization of new leads and is
preferable to the laborious target-specific lead identification for the R&D
sector of a pharma company.
b) Claims Databases for finding patients:

The Claims Database includes the use of applications such as APLD
(Anonymized Patient-Level Data) and Truven Marketscan for identifying
patients exhibiting characteristic symptoms correlating to the diagnosis code
available in the software, with this, an undiagnosed case can also be identified.
This approach can be utilized for finding orphan diseases, which are often
left undiagnosed. By applying ML, early disease progression can be identi-
fied, and pharma companies can utilize this information for creating orphan
drugs which are usually very expensive per patient revenue compared to the
cost allotted for drug discovery. Some examples of ML-based approaches
for identifying rare diseases are CART (Classification and Regression Tree)
models such as C5, standard decision trees, and random forests.
c) Patient medical history and treatment pathways:
A patient medical history reveals the journey through various treatment
and therapy approaches which can be further utilized to identify disease
- Relates genetic
information to
human functions
- High Throughput
Screening (HTS)
- Pharmacogenomics
- Quality Control - Prediction of 3D

and assurance structure of target
Application of protein
- Pharmacovigilance
Bioinformatics,
- Safety protocols AI, and ML in - Hit Identification
- Clinical studies Pharmacy field - Docking scores
- Market analysis - Lead Optimization
- Drug Activity
- ADMET studies
- Bioactivity
- Formulation
development
- Scale up
Figure 1.1 Applications of bioinformatics, AI, and ML in the pharmacy sector.

pathways and dosage regimens [52]. In this sector, machine learning is uti-
lized as means of avoiding data clustering from scoring models by creating
a time series data correlated with treatment pathways for faster processing
and improved efficiency, by incorporating database tools like kdb+, and
Tensorflow.
d) Enhancing Commercial market survey:

Physical methods as means of surveying physicians’ drug prescribing
patterns is often a tedious task for marketing individuals in the pharma
sector. These surveys are usually conducted for identifying dosage trends
that can be further utilized for new drug developments [52]. ML models,
such as Associative Rules Mining or “apriori”, make the laborious task
easier by providing quantitative variables related to Rx records, which
can then be applied for large data assessment. Figure 1.1 summarizes
some of the applications of bioinformatics, AI, and ML in the pharmacy
sector.
1.3.2 Limitations of ML
The major issue while using ML is obtaining accurate diagnostic values. In
ML, a predefined set of algorithms are set for a particular disease but it is
not generally quantitative as the human diseased state is due to innumera-
ble complex pathways going inside. Rather than quantitative assessment in
identifying a disease and designing the formulation for them, experience
and expertise are needed for diagnosis and dose requirement calculation.
Nevertheless, ML can be utilized for creating a huge dataset that can give
fairly quantified data for further usage [52].
1.4 Conclusion and Future Prospects

It is determined that a summary of this interdisciplinary subject is formed
by producing a unique precise understanding that is provided by the reac-
tion of biology and computer science, as well as certain evaluation aspects
such as statistics and mathematics, to end in the recently hatched field after
this powerful reaction. Bioinformatics has a promising outlook in many
biological and living fields, but one of the most critical difficulties that must
be addressed is the integration of a large and diverse set of data sources and
databases for the improvement of life and a massive biological change.
References
1. Quazi, S., Role of Artificial Intelligence and Machine Learning in Bioinformatics:
Drug Discovery and Drug Repurposing, 2021. https://www.preprints.org/
manuscript/202105.0346/v1
2. Selvaraj, C., Chandra, I., Singh, S.K., Artificial intelligence and machine
learning approaches for drug design: Challenges and opportunities for the
pharmaceutical industries. Mol. Divers., 3, 1893–1913, 2021.
3. Henstock, P.V., Artificial intelligence for pharma: Time for internal invest-
ment. Trends Pharmacol. Sci., 40, 8, 543–546, 2019.
4. Chen, C., Hou, J., Tanner, J.J., Cheng, J., Bioinformatics methods for mass
spectrometry-based proteomics data analysis. Int. J. Mol. Sci., 21, 8, 2873,
2020.
5. Can, T., Introduction to bioinformatics. Methods Mol. Biol., 1107, 51–71,
2014.
6. Najarian, K., Deriche, R., Kon, M.A., Hirata, N.S.T., Bioinformatics and bio-
medical informatics. Sci. World J., 2013, 591976, 2013.
7. Chakraborty, C., Doss, C.G.P., Zhu, H., Agoramoorthy, G., Rising strengths
Hong Kong SAR in bioinformatics. Interdiscip. Sci., 9, 2, 224–236, 2017.
8. Pallen, M.J., Microbial bioinformatics 2020. Microb. Biotechnol., 9, 5, 681–
686, 2016.
9. Wooller, S.K., Benstead-Hume, G., Chen, X., Ali, Y., Pearl, F.M.G.,
Bioinformatics in translational drug discovery. Biosci. Rep., 37, 4,
BSR20160180, 2017.
10. Pop, M. and Salzberg, S.L., Bioinformatics challenges of new sequencing
technology. Trends Genet., 24, 3, 142–149, 2008.
11. Tillett, R.L., Sevinsky, J.R., Hartley, P.D. et al., Genomic evidence for reinfec-
tion with SARS-CoV-2: A case study. Lancet Infect. Dis., 21, 1, 52–58, 2021.
12. Rothberg, J., Merriman, B., Higgs, G., Bioinformatics. Introduction. Yale J.
Biol. Med., 85, 3, 305–308, 2012.
13. Mbah, C.J. and Okorie, N.H., Pharmaceutical bioinformatics: Its relevance
to drug metabolism. Madridge J. Bioinform. Syst. Biol., 1, 1, 19–26, 2019.
[Internet] Available from: https://madridge.org/journal-of-bioinformatics-
and-systems-biology/mjbsb-1000104.php.
14. Bayat, A., Science, medicine, and the future: Bioinformatics. BMJ, 324, 7344,
1018–1022, 2002.
15. Altschul, S.F., Gish, W., Miller, W., Myers, E.W., Lipman, D.J., Basic local
alignment search tool. J. Mol. Biol., 215, 3, 403-410, 1990.
16. Davies, M., Nowotka, M., Papadatos, G. et al., ChEMBL web services:
Streamlining access to drug discovery data and utilities. Nucleic Acids Res.,
43, W1, W612–W620, 2015. [Internet] Available from: https://www.ebi.
ac.uk/chembl/.
17. Thorvaldsdottir, H., Robinson, J.T., Mesirov, J.P., Integrative genomics viewer
(IGV): High-performance genomics data visualization and exploration.
Brief. Bioinform., 14, 2, 178–192, 2013. [Internet] Available from: https://
academic.oup.com/bib/article-lookup/doi/10.1093/bib/bbs017.
18. Nascimento, A.C.A., Prudêncio, R.B.C., Costa, I.G., A multiple kernel
learning algorithm for drug-target interaction prediction. BMC Bioinf.,
17, 1, 46, 2016. [Internet] Available from: http://www.biomedcentral.
com/1471-2105/17/46.
19. Daina, A. and Zoete, V., Application of the SwissDrugDesign online resources
in virtual screening. Int. J. Mol. Sci., 20, 18, 4612, 2019.
20. UCSF Chimera. [Internet] Available from: https://www.rbvi.ucsf.edu/
chimera/.
21. Jumper, J., Evans, R., Pritzel, A. et al., Highly accurate protein structure
prediction with AlphaFold. Nature, 596, 7873, 583–589, 2021. [Internet]
Available from: https://www.nature.com/articles/s41586-021-03819-2.
22. Snijder, E.J., Decroly, E., Ziebuhr, J., The nonstructural proteins directing
coronavirus RNA synthesis and processing. Adv. Virus Res., 96, 59–126,
2016.
23. Mamoshina, P., Vieira, A., Putin, E., Zhavoronkov, A., Applications of
deep learning in biomedicine. Mol. Pharm., 13, 5, 1445–1454, 2016.
[Internet] Available from: https://pubs.acs.org/doi/10.1021/acs.molpharmaceut.
5b00982.
24. Schneider, P., Walters, W.P., Plowright, A.T., Sieroka, N., Listgarten, J.,
Goodnow, R.A. Jr, Fisher, J., Jansen, J.M., Duca, J.S., Rush, T.S., Zentgraf, M.,
Hill, J.E., Krutoholow, E., Kohler, M., Blaney, J., Funatsu, K., Luebkemann,
C., Schneider, G., Rethinking drug design in the artificial intelligence era.
Nat. Rev. Drug Discov., 19, 5, 353–364, 2020.
25. Agatonovic-Kustrin, S. and Beresford, R., Basic concepts of artificial neural
network (ANN) modeling and its application in pharmaceutical research.
J. Pharm. Biomed. Anal., 22, 5, 717–727, 2000.
26. Sakiyama, Y., The use of machine learning and nonlinear statistical tools for
ADME prediction. Expert Opin. Drug Metab. Toxicol., 5, 2, 149–169, 2009.
27. Gobburu, J.V.S. and Chen, E.P., Artificial neural networks as a novel approach
to integrated pharmacokinetic—Pharmacodynamic analysis. J. Pharm. Sci.,
85, 5, 505–510, 1996.
28. Merk, D., Friedrich, L., Grisoni, F., Schneider, G., De novo design of bioac-
tive small molecules by artificial intelligence. Mol. Inform., 37, 1–2, 1700153,
2018. [Internet] Available from: https://onlinelibrary.wiley.com/doi/10.1002/
minf.201700153.
29. Klopman, G., Chakravarti, S.K., Zhu, H., Ivanov, J.M., Saiakhov, R.D., ESP:
A method to predict toxicity and pharmacological properties of chemicals
using multiple MCASE databases. J. Chem. Inf. Comput. Sci., 44, 2, 704–715,
2004. [Internet] Available from: https://pubs.acs.org/doi/10.1021/ci030298n.
30. Hamet, P. and Tremblay, J., Artificial intelligence in medicine. Metabolism,

69, S36–S40, 2017. [Internet] Available from: https://linkinghub.elsevier.
com/retrieve/pii/S002604951730015X.
31. Bain, E.E., Shafner, L., Walling, D.P. et al., Use of a novel artificial intelligence
platform on mobile devices to assess dosing compliance in a phase 2 clinical
trial in subjects with schizophrenia. JMIR mHealth uHealth, 5, 2, e18, 2017.
[Internet] Available from: http://mhealth.jmir.org/2017/2/e18/.
32. Sharma, T., Mankoo, A., Sood, V., Artificial intelligence in advanced phar-
macy. Int. J. Sci. Res. Arch., 2, 1, 047–054, 2021. [Internet] Available from:
https://ijsra.net/content/artificial-intelligence-advanced-pharmacy.
33. Mundy, L., Trowman, R., Kearney, B., Improving access to high-cost technol-
ogies in the Asia region. Int. J. Technol. Assess. Health Care, 35, 3, 168–175,
2019.
34. Technologies, O., Automatic process control control (APCS). Access date
03/05/2022; https://rivs.ru/en/asu_tp#:~:text=Automated%20process%20
control%20system%20(APCS,mathematical%20support%20and%20
operational%20personnel.
35. Corsello, S.M., Bittker, J.A., Liu, Z. et al., The drug repurposing hub: A
next-generation drug library and information resource. Nat. Med., 23, 4,
405–408, 2017. [Internet] Available from: http://www.nature.com/articles/
nm.4306.
36. Mak, K.-K. and Pichika, M.R., Artificial intelligence in drug development:
Present status and future prospects. Drug Discovery Today, 24, 3, 773–780,
2019. [Internet] Available from: https://linkinghub.elsevier.com/retrieve/pii/
S1359644618300916.
37. Das, S., Dey, R., Nayak, A.K., Artificial intelligence in pharmacy. Indian J.
Pharm. Educ. Res., 55, 2, 304–318, 2021.
38. Del Rio-Bermudez, C., Medrano, I.H., Yebes, L., Poveda, J.L., Towards a sym-
biotic relationship between big data, artificial intelligence, and hospital phar-
macy. J. Pharm. Policy Pract., 13, 1, 75, 2020.
39. Weber, G.M., Mandl, K.D., Kohane, I.S., Finding the missing link for big bio-
medical data. JAMA, 24, 2479–2480, 2014.
40. Luo, Y., Thompson, W.K., Herr, T.M. et al., Natural language processing for
EHR-based pharmacovigilance: A structured review. Drug Saf., 40, 11, 1075–
1089, 2017.
41. Gottesman, O., Kuivaniemi, H., Tromp, G. et al., The electronic medical
records and genomics (eMERGE) network: Past, present, and future. Genet.
Med., 15, 10, 761–771, 2013.
42. Abdullah, Y.I., Schuman, J.S., Shabsigh, R., Caplan, A., Al-Aswad, L.A.,
Ethics of artificial intelligence in medicine and ophthalmology. Asia-Pac. J.
Ophthalmol., 10, 3, 289–298, 2021.
43. Trenfield, S.J., Awad, A., McCoubrey, L.E. et al., Advancing pharmacy and
healthcare with virtual digital technologies. Adv. Drug Deliv. Rev., 182,
114098, 2022.
44. Gunčar, G., Kukar, M., Notar, M. et al., An application of machine learning to
haematological diagnosis. Sci. Rep., 8, 1, 411, 2018. [Internet] Available from:
http://www.nature.com/articles/s41598-017-18564-8.
45. Koohy, H., The rise and fall of machine learning methods in biomedical
research. F1000Res., 6, 2012, 2018. [Internet] Available from: https://f1000re-
search.com/articles/6-2012/v2.
46. Young, J.D., Cai, C., Lu, X., Unsupervised deep learning reveals prognos-
tically relevant subtypes of glioblastoma. BMC Bioinf., 18, S11, 381, 2017.
[Internet] Available from: http://bmcbioinformatics.biomedcentral.com/
articles/10.1186/s12859-017-1798-2.
47. Chen, H., Engkvist, O., Wang, Y., Olivecrona, M., Blaschke, T., The rise of
deep learning in drug discovery. Drug Discovery Today, 23, 6, 1241–1250,
2018. [Internet] Available from: https://linkinghub.elsevier.com/retrieve/pii/
S1359644617303598.
48. Beneke, F. and Mackenrodt, M.-O., Artificial intelligence and collusion.
IIC-Int. Rev. Intellect. Prop. Compet. Law, 50, 1, 109–134, 2019. [Internet]
Available from: http://link.springer.com/10.1007/s40319-018-00773-x.
49. Tkachenko, N., Machine learning in healthcare: 12 real-world use cases to
know, in IEEE Access, 2021. [Internet] Available from: https://nix-united.
com/blog/machine-learning-in-healthcare-12-real-world-use-cases-to-
know/; Bharadwaj, H.K. et al., A review on the role of machine learning in
enabling IoT based healthcare applications. IEEE Access, 9, 38859–38890,
2021.
50. Vamathevan, J., Clark, D., Czodrowski, P. et al., Applications of machine
learning in drug discovery and development. Nat. Rev. Drug Discovery, 18, 6,
463–477, 2019.
51. Mamoshina, P., Volosnikova, M., Ozerov, I.V. et al., Machine learning on
human muscle transcriptomic data for biomarker discovery and tissue-
specific drug target identification. Front. Genet., 242, 9, 2018.
52. Dasgupta, N., Real-world use cases for AI & ML in pharma. Access date
03/05/2022; https://www.rxdatascience.com/blog/top-use-cases-for-machine-
learning-in-pharma
53. Yuan, Y., Pei, J., Lai, L., Ligbuilder 2: A practical de novo drug design
approach. J. Chem. Inf. Model., 51, 5, 1083–1091, 2011. [Internet] Available
from: https://pubs.acs.org/doi/10.1021/ci100350u.
54. Zhu, T., Cao, S., Su, P.-C. et al., Hit identification and optimization in virtual
screening: Practical recommendations based on a critical literature analysis.
J. Med. Chem., 56, 17, 6560–6572, 2013. [Internet] Available from: https://
pubs.acs.org/doi/10.1021/jm301916b.
55. Anderson, A.C., Structure-based functional design of drugs: From tar-
get to lead compound. Methods Mol. Biol., 2012, 359–366, 2012. [Internet]
Available from: http://link.springer.com/10.1007/978-1-60327-216-2_23.
2
Artificial Intelligence and Machine
Learning-Based New Drug Discovery
Process with Molecular Modelling
Isha Rani1, Kavita Munjal2, Rajeev K. Singla3,4 and Rupesh K. Gautam5*
Spurthy College of Pharmacy, Marasur Gate, Bengaluru, Karnataka, India

1
2
MM College of Pharmacy, MM (Deemed to be) University-Mullana,
Ambala, Haryana, India
3
Institutes for Systems Genetics, Frontiers Science Center for Disease-Related
Molecular Network, West China Hospital, Sichuan University, Chengdu, China
4
iGlobal Research and Publishing Foundation, New Delhi, India
5
Department of Pharmacology, Indore Institute of Pharmacy, IIST Campus,
Rau, Indore (M.P.), India
Abstract
Drug development is a time-consuming, expensive and extremely risky proce-
dure. Up to 90% of drug concepts are discarded due to challenges such as safety,
efficacy and toxicity resulting in significant losses for the investor. The use of arti-
ficial intelligence (AI), namely machine learning and deep learning algorithms,
to improve the drug discovery process is one technique that has arisen in recent
years. AI has been effectively used in drug discovery and design. This chapter
includes these machine learning approaches in depth, as well as their applications
in medicinal chemistry. The current state-of-the-art of AI supported pharmaceu-
tical discovery is discussed, including applications in structure and ligand-based
virtual screening, de novo drug design, drug repurposing, and factors related, after
introducing the basic principles, along with some application notes, of the various
machine learning algorithms. Finally, obstacles and limits are outlined, with an
eye towards possible future avenues for AI-supported drug discovery and design.
Keywords: Artificial intelligence, drug development, drug discovery, lead

optimization, molecular modelling, virtual screening, de novo drug design, drug
repurposing
*Corresponding author: drrupeshgautam@gmail.com
Vivek Chavda, Krishnan Anand and Vasso Apostolopoulos (eds.) Bioinformatics Tools for Pharmaceutical
Drug Product Development, (19–36) © 2023 Scrivener Publishing LLC
19
Abbreviations
AI Artificial intelligence
RNA Ribonucleic acid
R & D Research and develoment
ML Machine Learning
SBVS Structure-based virtual screening
VS Virtual screening
CADD Computer aided drug design
PDB Protein data bank
SAS Synthetic accessibility score
2.1 Introduction
The development of pharmaceutical drugs is a time-consuming and costly
process. Pharmaceutical and biotechnology companies often spend over
$1 billion to develop a drug to the market, and can take anywhere from
10 to 20 years. This process is extremely risky with up to 90% of new drug
concepts are discarded due to difficulties such as safety and efficacy, result-
ing in significant loss for the investor [1, 2]. Traditional drug discovery
methods are target-driven, in which a known target is used to screen for
small molecules that either interact with it or affect its function in cells.
These approaches work well for easily druggable targets with well-defined
structures and well-understood cellular interactions. However, due to the
complex nature of cellular interactions and limited knowledge of intrica-
cies, these methods are severely limited [1, 3].
The term “artificial intelligence” (AI) refers to intelligence displayed by
computers. When a computer exhibits cognitive behavior similar to that
of humans, such as learning or problem solving, this term is employed [4].
AI makes use of systems and software that can read and learn from data
in order to make independent judgments in order to achieve certain goals.
Machine learning, for example, is a well-established technology for learning
and predicting novel features [5]. By finding novel relationships and infer-
ring the functional importance of distinct components of a biological path-
way, AI can overcome these obstacles. Complex algorithms and machine
learning are employed by AI to extract useful information from enormous
datasets. As such, a dataset of RNA sequencing can be used to discover
genes whose expression correlates with a specific biological situation. AI
can also be used to discover compounds that could bind to ‘undrugga-
ble targets,’ or proteins with unknown structures. A predicted collection
of compounds may be easily identified in a relatively short length of time
AI and ML in Drug Discovery Process 21
by iterative simulations of different compounds’ interactions with tiny por-

tions of a protein [6]. Companies that are commercializing AI drug discov-
ery platforms and AI-discovered pharmaceuticals have demonstrated that
using algorithms can reduce the multi-year process down to a few months
[7]. This large reduction in development time, as well as the quantity of
compounds that must be manufactured for laboratory testing provides for
significant cost savings, addressing two key challenges in pharmaceutical
R&D. The drug development process covers mainly virtual screening, de
novo drug design, lead optimization (predicting and optimizing compound
properties), planning chemical synthesis, pre-clinical testing, translation
to human clinical trials, as well as manufacturing processes, scale up etc.
All of these procedures add to the difficulty of identifying effective drugs
to treat an illness. As a result, the most important concern facing pharma-
ceutical businesses is how to manage the process’s cost and pace [8–10].
All of these issues need to be answered in a simple and scientific man-
ner, reducing the time and expense of the procedure. Furthermore, the rise
in data digitization in the pharmaceutical and healthcare industries stimu-
lates the use of AI to solve the issues of analyzing complicated data [11, 12].
This chapter will comprise the role of AI in drug discovery with respect
to virtual screening (VS), de novo design, synthesis planning, quality con-
trol and quality assurance etc. Various AI applications in drug design and
discovery, including primary and secondary screening, drug toxicity, drug
pharmacokinetics, medication dose efficacy, drug repositioning, poly-
pharmacology, and drug-target interactions etc. all of which are presented
herein.
2.2 Artificial Intelligence in Drug Discovery

The drug industry has seen a significant surge in knowledge digitization
in recent years. Using an AI algorithm in drug development is a step-by-
step procedure that begins with an identification of the problem [13]. This
method includes processes such as problem recognition, AI architecture
design, analysis of data, model assessment and evaluation, and understand-
ing and presenting the data. To be more particular, prior to making any
specific design decisions, one must have a thorough understanding of the
problem (step 1). With the goal of AI programming in sight, the next stage
is to create a model architecture that is suitable for this project (step 2). This
stage entails selecting an appropriate algorithm and determining appropri-
ate set of parameters initial values. Once settling on a tentative design, data
is collected (step 3 in Figure 2.1). The quality, amount, and generalizability
Building of an AI planner
Model
Problem AI arhitecture Analysis of
assessment
recognition design data
and evaluation
*Distinction *Fixing *Data pre- *Training

(Active initial values processing algorithm
(stochastic
and inactive for AI *Preparation
gradient
molecules) algorithm of training descent, etc.)
*Creating *Create a and test data *Evaluation
new model *Quality of mechanism
molecule architecture data (bootstraping,
Cross Validation)
Figure 2.1 The process of constructing an AI planner in general.
of initial data have a significant effect on the performance of an AI sys-

tem. After the basic architecture and data repositories have been developed,
learning algorithm and validation begins (step 4). The goal of the training
process is to determine a set of conditions that will limit the forecast error.
The completed AI model is expected to articulate the underlying relation-
ship among both molecular descriptions and professional goals [14].
2.2.1 Training Dataset Used in Medicinal Chemistry

A constant input string is the most commonly utilized input data format
in drug development (e.g. bitstrings, real-valued digits vector) [15]. Many
AI-assisted drug development tools produce numerical values as output
Table 2.1 Input and output data patterns for the development of AI algorithms
in drug discovery.
Data format
Input X • Subsequent data (e.g. SMILES String, data from time period)
• Structure of macromolecules structures, crystal structure of
molecules, retrieving receptor-ligand interaction vector with
a predefined size (bitstrings, real-valued digits vector)
Output Y • Binary numbers for binary classification tasks
• Numeric values for cluster analysis,
• Legitimate numbers for regression issues
• Subsequent data (e.g. SMILES String, Series of amino acids)
• Solitary learning relates to a single input column;
multitasking process with respect to the many data columns
data, such as binary numbers for binary classification tasks, numeric values
for cluster analysis, and legitimate numbers for regression issues, which
typically incorporate genuine biological outcomes [16]. Table 2.1 rep-
resents various data patterns used as input and output for the development
of AI algorithms in drug discovery.
2.2.2 Availability and Quality of Initial Data

It is necessary to give extra attention to the original data’s predictive value
and attribute values. To commence, evaluate if the efficiency of the train-
ing set is acceptable or not (Figure 2.2). If it is satisfactory on training set
then the performance of test set need to be checked. If the training set’s
result are not up to par, alterations to the AI architecture is required [17].
Collecting more data will be one of the most effective methods if the differ-
ence between training and test-set effectiveness is unacceptable. To acquire
an understanding of the model’s representativeness and stability, func-
tionality the input data and retrain the model with varying proportions
of every stratum selected evenly, as well as arbitrarily. Almost usually, it is
desirable to obtain good-quality raw data at a lower cost. A more common
Yes Is the training set’s No

effectiveness satisfactory?
Yes Is the effectiveness of No Do some changes

test set satisfactory? in AI architecture
(finer tweaking
of some controlled
parameters as batch
size and learning
speed)
Developed Need to Improve
sustainable Learning algorithm
model
Is
Yes performance on No
training set
better?
Need to Improve
Learning algorithm
Figure 2.2 General flowchart to describe model optimization techniques.

case is that such procedures are prohibitively expensive or impossible to

implement, as is frequently the case in medication development using bio-
logical information [18, 19]. To enhance the model with knowledge about
related data, the training set can be supplemented by creating multitask-
ing or transferable learning processes [20]. Based on the problem at mind,
the simulation can be enhanced by hyper-parameter tuning, regularization
or the use of additional specialized neural net topologies. If the test dataset
efficiency is still significantly lower than the training sample after tweak-
ing the regularization hyperparameters, and collecting additional training
data is impractical, the simplest solution to reduce the generalization error
may be to enhance the deep learning model itself. The GAN method, for
example, may be used to efficiently construct artificial molecule structures
with desired attributes from both restricted labelled and unlabeled input.
2.3 AI in Virtual Screening

The main goal of the drug development process is to find bioactive molecules
that can help in disease therapy. The process begins with the discovery of
molecular targets for a specific drug (natural or synthetic) and ends with the
confirmation of those targets. The drug development procedure is lengthy
and very costly. As stated above, drug discovery journey goes through all
phases of new drug development, from target identification through drug
registration. Moreover, the creation of a new medicines/drugs can costs
between $1 and $2 billion USD on average and can take 10–20 years for its
development [21–23]. Due to such problematic issues, researchers are con-
tinually investing in the development of novel ways to improve the drug dis-
covery process’ efficiency. Computer-aided drug design is one of the most
widely utilized methods for reducing medication development costs and
time (CADD). CADD approaches the molecular modeling technique and
enables for improved experiment focus, cutting down on the time and cost
of medication development [24]. Molecular modeling allows researchers to
examine a large number of molecules in a short amount of time, illustrating
how they interact with pharmacological targets even before they are synthe-
sized. It starts out prediction of various important characteristics of mole-
cules such as toxicity, activity, bioavailability, and efficacy etc. and then after
validation and proper interpretation of above stated parameters, it moves
further for in vitro and in vivo testing. Thus, by following such techniques,
research goes with proper planning and direction [25–27].
Structure-based virtual screening (SBVS), one of the most promising in
silico tools for drug research, is resilient and useful in this context. Virtual
screening (VS) can help with hit discovery (finding active drug candidates)
and lead optimization (transforming biologically active compounds into
appropriate pharmaceuticals by increasing their physicochemical qual-
ities). Finally, these improved leads go through preclinical and clinical
testing before being approved by regulatory agencies [28]. For VS, large
databases of known 3D structures are analyzed automatically utilizing
computational approaches. Initially, thousands of possible active ligands
are screened for a target. The probable ligands are selected using VS with
AutoDock Vina. DOCK 6 is then used for carrying out molecular dynam-
ics simulations. By this way, active compounds with promising results are
selected for further experimental testing. This leads to the most promis-
ing synthesized molecules. VS techniques do even wonders by identifying
if any compound is toxic. It gives proper ADMET (absorption, distribu-
tion, metabolism, excretion and toxicity) analysis of tested molecule. At
the binding site, search algorithms are utilized to systematically evaluate
ligand orientations and conformations. In rigid docking, the search algo-
rithms make use of translational and rotational degrees of freedom to
explore alternative positions of ligands at the active binding site, whereas
in flexible docking, conformational degrees of freedom are added to the
translations and rotations of the ligands. Search algorithms use a variety
of strategies to anticipate the correct conformation of ligands, including
checking the chemistry and geometry of the atoms involved (DOCK 6,
FLEXX and genetic algorithm) [29–31]. Further, SBVS uses scoring func-
tions to evaluate the force of non-covalent contacts between a ligand and a
molecular target, and it tries to anticipate the optimum interaction mode
between two molecules to form a stable complex [32].
The availability of a 3D structure of the target protein and the ligands to
be docked is a need for executing VS [33, 34]. Some databases have been
built to store 3D molecular structures, such as, Protein Data Bank (PDB),
PubChem, ChemSpider, Brazilian Malaria Molecular Targets (BraMMT),
Drugbank [35].
2.4 AI for De Novo Design

The purpose of de novo design is to find new bioactive molecules that can meet
a variety of important performance parameters, including as activities, selec-
tion, physiochemical characteristics, and ADMET qualities, all at the same
time [36]. Numerous approaches and software solutions have been intro-
duced [37]. But de novo design has not seen a widespread use in drug discov-
ery. This is at least partially related to the generation of compounds, which are
synthetically difficult to access. The field has seen some revival recently due
to developments in the field of artificial intelligence. An interesting approach
is the variational autoencoder, which consists of two neural networks, an
encoder network and a decoder network [38]. The encoder network converts
the chemical constituents described by the SMILES description into a legit-
imate constant vector. The decoding section can convert variables from the
latent space into active compounds. An in-silico model employed this feature
to search for best solution in subspace, and the decoder networks used this
feature to reverse translate such matrices into actual molecules. For the major-
ity of reverse translates, one molecule predominate, while minor structural
changes occur with a lower likelihood. The researchers trained a framework
on the QED drug-likeness rating [39] and the synthetic accessibility score
SAS [40] using the latent feature model. It might be possible to create a track
of compounds with specific targeted qualities. A generating model generates
unique chemical structures in the antagonistic learning algorithm. While the
predictive model tries to mislead the discriminative model, a second dis-
criminative adversarial model is trained to distinguish genuine molecules
from produced ones. In generating mode, the antagonistic algorithm created
considerably more acceptable architectures than the variational learning
algorithm. Novel structures anticipated to be effective in opposition to the
dopamine receptor type 2 might be produced using an in-silico approach.
A generative adversarial network (GAN) was utilized by Kadurin et al.
to suggest chemicals with suspected anticancer characteristics [41].
Numerous distinct architectures have just been constructed, each of which
is possible to produce complete and effective new structures one of them is,
Recursive neural networks (RNN) [42]. These approaches can be used to
explore a novel chemical space, with the created molecules’ property distri-
butions being similar to the training space. The methodology’s initial poten-
tial application was successful, with four out of five molecules exhibiting
the anticipated activity. However, further expertise with the vastness of the
chemical space examined and chemical validity of the defined compounds
is required.
2.5 AI for Synthesis Planning

Molecule synthesis is one of the most difficult problems in organic chem-
istry, and a chemist’s conventional method to solving a problem is based
on experience and is a repeated, time-consuming operation that frequently
results in non-optimized solutions [43, 44]. Artificial intelligence and
machine learning have shown their potential usefulness in small molecule
predictive chemistry and synthetic planning; at least a few businesses have

reported using in silico synthetic planning into their overall approach to
obtaining target molecules. Despite the fact that chemistry has a reputation
for being a conservative science that is hesitant to adopt new ideas, AI is
rapidly being researched across chemical disciplines [43]. Computational
approaches have long been used in medicinal chemistry, supplemented by
computer-aided drug design and cheminformatics, to aid in the search for
and optimization of active molecules [45]. The synthesis planning is a key
element in the drug discovery process. New computational approaches such
as, prediction of outcomes of a reaction based on given set of inputs, pre-
diction of yield of chemical reactions, and, involvement of retrosynthetic
planning are utilized. Knowledge-based systems based on expert-derived
rules or rules directly derived from reaction datasets dominate retrosyn-
thetic planning. According to a recent study, a variety of computer strat-
egies were utilized to forecast forward synthesis. They rate routes based
on the retro synthesis technique. In one technique, chemical descriptors
from quantum mechanics are merged with manually programmed rules
and machine learning to anticipate the reaction and its products at the
conclusion. This approach or process is used to forecast the outcome of a
multi-step analytical response [46].
2.6 AI in Quality Control and Quality Assurance

The balancing of numerous factors is required to manufacture the required
product from raw materials [47]. Manual intervention is required for product
quality control tests and also batch-to-batch uniformity. This highlights the
necessity for AI implementation at this level [48]. The Food and drug admin-
istration changed the Current Good Manufacturing Practices (cGMP) by
proposing a “Quality by Design” plan to better understand the important
operations and specifications that determine the final quality of pharmaceu-
tical products [49]. Gams et al. employed a mix of human and AI efforts to
examine early data from manufacturing batches and create regression trees.
Those were then turned into principles, which the personnel assessed in
order to lead the manufacturing cycle [47]. With the use of ANN, Goh et al.
investigated the dissolution rate, a measure of batch-to-batch uniformity of
some drugs [50]. AI could be used to regulate in-line production processes
in order to reach the target product standard [49]. The freeze-drying process
is monitored using an ANN that employs a mixture of self-adaptive pro-
gression, local search, and backpropagation methods. This could be utilized
to estimate temperature and desiccated-cake thickness at a later time point
Another random document with
no related content on Scribd:
valtaamisen tärkeyttä. Ja samaan aikaan riensi Kerenski rintamalle
kiihoittamaan kurittomia ja saksalaisten kanssa jo veljeilyn alkaneita
sotalaumoja mielettömään hyökkäykseen. Kerenskin puhelahja
tällöin vielä tepsi ja onneton hyökkäys saksalaisia vastaan alkoikin
kesäkuussa v. 1917 tuottaen hirvittäviä tappioita ja huomattavasti
lähentäen Kerenskin hallituksen kukistumista.
Entä, miten suhtautui hallitus pitkästä unestaan heräävään

talonpoikaisväestöön, tuohon kansan enemmistöön, jota ei koskaan
oltu otettu huomioon valtiollisena tekijänä? Älysikö se, että silläkin oli
toiveita ja odotuksia, jotka täytyisi täyttää? Vielä mitä. Kerenskin
hallitus katsoi maakysymyksen niin monimutkaiseksi, että sen voi
ratkaista tyydyttävästi vasta perustava kokous, jonka
kokoonkutsumista luvattiin. Sillä aikaa talonpojat, joita ilman
maaliskuun vallankumous oli suoritettu, puuhailivat omalaatuisissa
vallankumouksellisissa hommissaan: ryöstelivät herraskartanolta ja
jakelivat kaikessa hiljaisuudessa maita tyydyttäen näin kauan
kytenyttä kostonhaluaan.
*****
Kerenskin monien vaihdosten alaisena olleitten hallitusten

viivytellessä näin suurten kysymysten ratkaisua ja tukehtuessa
entisten tsaarihallitusten tavoin äärettömiin paperimassoihin johtivat
bolshevikit Työläis-, talonpoikais- ja sotilasneuvostoissa ripeästi
omaa politiikkaansa. Nämä neuvostot olivat syntyneet jonkunlaiseksi
parlamentiksi Valtakunnan duuman sijaan. Alussa toimivat ne
veljellisessä sovussa hallituksen kanssa, pienemmät erimielisyydet
kykeni Kerenskin kaunopuheisuus aina sovittamaan. Mutta syksyllä
v. 1917 oli Lenin ynnä muita bolshevistisia johtajia ilmaantunut
poliittiselle näyttämölle ja heidän vaikutuksensa alkoi piankin tuntua
neuvostoissa. Niiden ja hallituksen välinen juopa syveni
syvenemistään ja lopuksi oli niissä varma bolshevistinen aines
enemmistönä. Hallituksen puuhia arvosteltiin ankarasti, sen
luottamusta kansan silmissä horjutettiin kaikin tavoin ja bolshevistista
oppia kylvettiin milloin julkisemmin, milloin salaisemmin.
Neuvostoissa se alkoi kyteä rauhanajatuskin. Jo heinäkuussa,

suurhyökkäyksen epäonnistuttua, olivat bolshevikit uskaltautuneet
kadulle, mutta silloin heidän mielenosoituksensa hallituksen toimesta
armotta hajoitettiin. Tarkemmin asioihin perehtyneestä näytti
kuitenkin silloin jo varmalta, että bolshevikien seuraava esiintyminen,
joka epäilemättä ennen pitkää tulisi tapahtumaan, oli oleva
hallitukselle musertava.
Pian kohotettiin neuvostoissa julki tunnuslause: kaikki valta

neuvostoille! Se oli bolshevikien taisteluhuuto ja -haaste Kerenskin
hallitukselle. Viimemainittu oli haparoituaan sinne ja tänne yrittänyt
yhdennellätoista hetkellä turvautua byrokratismiinkin, mutta
verrattain huonolla menestyksellä. Kerenski itse, entinen
vallankumouksen sankari, oli helppo leimata kansan asian pettäjäksi.
Hän oli mahdollisesti uhkaavaa loppua aavistaessaan, koettanut
pukeutua diktaattorin viittaan, mutta se ei hänelle sopinut. Koko
yritys jäi vain huonon näyttelijän turhaksi ponnistukseksi esittää
kuningasta, vaikka hänelle paremmin sopisi hovimestarin osa.
Bolshevikit osasivat taitavasti käyttää hyväkseen kaikkea sitä,

minkä Kerenski oli laiminlyönyt. He osoittivat sodan jatkamisen
turhuuden, vetosivat rauhaa janoavaan kansaan. Tähän
vetoomukseen vastasi molempien pääkaupunkien työväestö ja
suurin osa armeijaa antamalla luottamuksensa heille. Bolshevikit
lupasivat ratkaista maakysymyksen talonpoikain eduksi heti, kun
porvarillinen hallitus oli kukistettu. Sisäläänien talonpoika tuskin tästä
lupauksesta tiesikään, mutta hänen sotapalveluksessa parhaillaan
oleva poikansa innostui ikihyväksi: pääseehän hän kotiin (kun
samalla sotakin loppuu) ja saadaan maata mielinmäärin. Bolshevikit
laskivat, että talonpojasta saadaan myöhemmin vahva tuki,
vallankumousta (tai kaappausta) tehtäessä häntä ei ole tarvittu
ennen, eikä tarvita nytkään. Sellaiset on suoritettu ennenkin vain
pääkaupungeissa, muu Venäjä laahaa kyllä aikanaan perässä.
Niin oli vähitellen bolshevikien valta neuvostoissa vahvistunut siinä

määrin, että he uskalsivat lähteä edellämainitut kaksi päälupausta:
rauha ja maata talonpojille, lippuunsa piirtäen vallananastuksen
tielle. Neuvosto otti itselleen vallan määrätä sotilasasioissa ja
marraskuun 7 päivän yönä miehitettiin tärkeimmät virastot. Jonkun
verran taisteluntapaista oli Pietarissa ainoastaan talvipalatsin
edustalla. Viime mainittuun oli sulkeutunut joukko junkkarikoululaisia,
jotka oli sieltä asein ulosajettava. Muuten meni kaikki niin hiljaisesti,
että pietarilaiset vasta sanomalehdistä näkivät saaneensa uuden
hallituksen. Moskovassa oli hiukan suurempi jähinä, mutta sangen
vähillä uhreilla sielläkin päästiin. Pietarin edustalla tekivät Matsinaan
paenneelle Kerenskille uskolliset kasakkajoukot vastarintaa, mutta
heidät masennettiin pian. Niin sekasortoista oli taistelu, että erään
mukanaolleen kertomuksen mukaan sama tykkipatteri oli kolme
kertaa taistellut kummallakin puolella. Ylpeänä saattoi tuleva
sotamarsalkka Trotski sähköttää ensimäisenä radiotietonaan
"kaikille, kaikille, kaikille", että "Työläis- ja talonpoikaisdemokratia" on
voittanut vihollisensa. Tuonnempana saanemme nähdä, minkälainen
"demokratia" tästä lapsesta varttui.
Työläisjoukot suurimmalta osaltaan riemuitsivat, Pietarissa

majailevat sotaväenosastot samoin, mutta sivistyneistö ennusti
uuden vallan sortuvan viiden päivän kuluessa. Muutamilla
bolshevikijohtajillakin tuntui olleen epäilyksiä, koskapa Pietarin koko
bolshevikien valtakauden aikainen diktaattori Zinovjev kuuluu
vallankaappauksesta keskusteltaessa huudahtaneen: "jos te voitte
taata, että pysymme vallassa kaksitoista tuntia, niin minä tulen
mukaan". Mutta tämä Zinovjev onkin demagogi Kerenskin malliin,
heikkotahtoinen ja häilyvä, eikä kaukonäköinen ja harkitseva kuin
Lenin, tai niinkuin lujatahtoinen ja järkkymätön Trotski.
Täytyy antaa bolshevikeille se tunnustus, että heidän ensi

otteensa olivat lupausten mukaisia. Heti paikalla julkaistiin dekreetit
rauhasta ja kaiken maan yhteiskunnallistuttamisesta. Tietysti ei
rauhaa solmittu yhdellä dekreetillä, vastustajallakin oli siinä asiassa
jotain sanottavaa, mutta sotaan kyllästyneiden mielet rauhoitettiin,
semminkin kun uusi hallitus ei vitkastellut antaa tietoja rauhanasian
edelleen kehittymisestä. Maaseudulla levisi tieto maakysymyksen
ratkaisusta herättäen siellä sekä kummastusta että mielihyvää.
Talonpojille oli yhteiskunnallistuttaminen outo käsite, mutta kokemus
vähitellen, jopa verrattain ripeässä tempossa, opetti heidät
ymmärtämään maakysymyksen omalla tavallaan. Tilanomistajat
ajettiin mailtaan muille markkinoille ja talonpojat ottivat maata niin
paljo kuin tarvitsivat.
Uusi valta konstruoitui nopeasti. Bolshevikien tunnuslause oli

toteutunut, kaikki valta oli nyt neuvostoilla. Nämä olivat suuria, sillä
valitsijayksikköjä oli paljo. Sellaisen "parlamentin" oli vaikea hallita.
Senpävuoksi luovuttikin se valtansa, s.t.s. toimeenpanevan vallan,
tuotapikaa kokoonkyhätylle ministeriölle, kansankomissaarien
neuvostolle.
Kerenskin aikaisesta ministeriöstä ei kansankomissaarien
neuvostoa oikeastaan eroittanut muu kuin nimi. Nimittämisperustekin
oli sama. Valtakunnan duuma, eräänlaisilla parlamentaarisilla
vaaleilla valittu edustuslaitos, oli jo aikaisemmin menettänyt
valtansa. Työläis-, talonpoikais- ja sotilasneuvostot hoitelivat entisen
duuman tehtäviä. Jo nimestäkin päättäen on viimemainitulla
luokkaluonne, niihin ei ollut tarkoitus laskea porvareita. Neuvostoissa
siis jo alusta alkaen piili diktatuurin siemen sellaiselle vallanpitäjälle,
joka pyrki sitä käyttämään.
Jo ennen bolshevikikumousta valmistuneessa (kirja tosin ilmestyi

vasta myöhemmin, mutta on päivätty syksyllä 1917) kirjassaan
"Valtio ja vallankumous" Lenin puhui köyhälistön diktatuurin
välttämättömyydestä "kommunismin ensi asteella". Tästä päättäen
olisi siis bolshevikeilla alunpitäen ollut tarkoitus saattaa valtiollinen
diktatuuri voimaan. Kuitenkin puhui Trotski, kuten edellä on kerrottu,
työläisten ja talonpoikain demokratiasta. Vaikeata on tällä haavaa
käytettävissä olevien lähteiden niukkuuden takia päättää, mitä
valtiollista järjestelmää voittajat aikoivat sovelluttaa. Mutta
bolshevikien vihollisten luku oli jo alunperin legio, jonka vuoksi
heidän kaikessa tapauksessa oli vihdoin soluttava leppymättömän
diktatuurin, jopa terrorinkin tielle.
III.
Bolshevikien taistelu vastavallankumousta ja keinottelua vastaan v.

1918 ensi puoliskolla.
Alussa todella näytti siltä, ettei boshevistinen kumous tulee

aiheuttamaan kerenskiläiseen hallitusjärjestelmään suuriakaan
oleellisia muutoksia. Neuvostoihin oli jo tavallaan totuttu, niitähän oli
viime aikoina siinnyt jokaiseen kolkkaan, yksinpä sotilasosastoihin
pienimpiä yksikköjä myöten. Sitäpaitsi oli niillä, jotka katsoivat
syrjäsilmin uutta valtaa, toivoa saada omia miehiään sijoitetuiksi
neuvostoihin. Olihan varsinkin armeijassa mitä kirjavinta ainesta ja
sanavalmius oli edelleenkin vaalikelpoisuusehdoista parhain. Löytyi
siis vastapuolellakin mahdollisuuksia miehittää neuvostot. Bolshevikit
eivät myöskään alussa osoittaneet ankaruutta vastustajiaan kohtaan.
Väliaikaisen (Kerenskin) hallituksen jäsenet jäivät paria poikkeusta
lukuunottamatta vapaalle jalalle, korkeita sotilashenkilöltä ei liioin
ahdistettu, m.m. Kerenskin esikuntapäällikkö kenraali Krasnov, joka
antautui Hatsinassa, pääsi melkein oitis vapaaksi. Näin näytti kaikki
kulkevan vanhaa latuaan.
Sen lisäksi oli vastapuolella yksi suuri toivo: Perustava kokous.
Alussa uskottiin vahvasti, että bolshevikit eivät tule häiritsemään sen
valitsemista, enemmän kuin kokoontumistakaan, ja oltiin lujasti
vakuutettuja sen antibolshevistisesta enemmistöstä. Vaalit saivatkin
tapahtua esteettömästi.. Niitten tulos: runsaasti bolshevikeja, mutta
kaikilla muilla puolueilla yhteensä huomattava äänten enemmistö.
Kun sitten valitut edustajat vuoden vaihteessa (1917—18) saapuivat
Pietariin ja istuntojen piti alkaa, hajoitti hallituksen määräämä
komissaari Uritski Perustavan kokouksen väkivallalla.
Hajoittamistilaisuudessa esiintyi muuan pääasiallisesti upseereista

kokoonpantu ja huonosti aseistettu joukko kokouksen puolesta.
Silloin käytti uusi valta ensi kerran järjestymisensä jälkeen asetta
vihollisiaan vastaan, samalla kun se Perustavan kokouksen
hajoittamisella ensi kerran paljasti diktatoorisen luonteensa.
Tästä hetkestä alkoi bolshevistinen hallitus epäluulolla katsella

kaikkia niitä, jotka eivät olleet vannoutuneet sen kannattajiksi.
Porvarilliset ja sosialistiset lehdet saivat tosin vielä ilmestyä, mutta
jälkisensuuri — ennakkosensuuria ei ollut — toimitti kiivaimmin
esiintyvät pois päiviltä. Lakkautetun sanomalehden sijaan voi kyllä
perustaa uuden, joten näin nimiä vaihdellen saattoivat lehdet
ilmestyä melkein keskeytymättä siihen saakka, kunnes ne kesällä v.
1918 kaikki hallituksen määräyksellä lopetettiin ja niiden
ilmestyminen kokonaan kiellettiin.
Venäjän sydämessä alkoivat sinne kokoontuneet kenraalit keräillä

ympärilleen valkoisia joukkoja ja esiintyä toinen toisensa perään,
väliin yhdessäkin, uusia vallanpitäjiä vastaan. Pääkaupungissakin
paljastui jokunen heikosti järjestetty salaliitto. Ententen kulta
viekoitteli pauloihinsa köyhtyneitä aristokraatteja ja virattomia
upseereja. Näin syntyi vehkeilyjä vähän siellä ja täällä ja
bolshevikien vasta perustama uusi orgaani tsheka (virallisesti:
Komitea taistelua varten vastavallankumousta ja keinottelua
vastaan) ponnisti voimansa äärimmilleen.
Paljon oli kevättalvella ja keväällä v. 1918 väkeä tshekan

vankiloissa Pietarissa. Bolshevikihallitus oli saanut solmituksi rauhan
Saksan kanssa ja tämä toimenpide, niin tyynnyttäväsi kuin se oli
vaikuttanutkin työläisiin ja maalaisväestöön (viimemainittuun
maakysymyksen talonpojille edullisen ratkaisun ohella), oli
syventänyt juopaa hallituksen ja sotahullujen upseerien välillä.
Ententen kiihoitus ja kulta löysivät jälkimäisten joukosta kiitollisen
maaperän ja antibolshevistinen propaganda huomattavasti vahvistui.
Luonnollisesti korjasi myös tsheka sitä runsaamman sadon.
Hallitus oli yhä lisääntyvien vihollistensa painostamana ja

mahdollisesti myös Suomen valkoisten hyökkäystä peläten siirtynyt
Moskovaan vieden sinne mukanaan yleisvenäläisen tshekan
päällikön Dsershinskin. Pietarin ja ympäristön
vastavallankumouksellisten kurissapitäminen jäi komissaari Uritskin
tehtäväksi.
Pietarin tsheka oli sijoittunut entiseen kaupunginpäällikön

virastoon Gorohovaja-kadulla. Pihan puoleinen sivusta oli järjestetty
pidätettyjen säilytyshuoneiksi, kadun puoleisissa rakennuksissa
majailivat tutkintatuomarit ja konttorihenkilökunta.
Kun minut eräänä päivänä kevättalvella useinmainittuna vuonna

tuupattiin muutamaan Gorohovajan säilytyshuoneeseen, tunsin heti,
että nyt saan uudistaa entisiä muistoja useista tsaarinaikuisista
vankiloista, joiden kanssa minulla oli ollut tilaisuus tehdä tuttavuutta
jokunen vuosi aikaisemmin. Hallitukset olivat vaihdelleet, kokonaisia
järjestelmiä sortunut, mutta vankilakulttuuri kukoisti ennallaan. Sama
ummehtunut, "ryssänhajun" kyllästyttämä ilma hyökkäsi heti nenään,
samalla tapaa kuin ennenkin otti starosta — vankien omasta
keskuudestaan valitsema yhteiskopin esimies — tulokkaan vastaan,
samat kysymykset: mikä nimesi, kuinka vanha, mistä syystä
vangittu. Nyt oli huomattavissa ainoastaan sellainen eroitus, että
melkein jokaisen nimen kohdalle oli merkitty vangitsemisen syyksi: ei
tiedä. Vangitkin olivat samallaisia kuin ennenkin, ainakaan ei ensi
katsannolla huomannut muuta eroa, kuin että upseeritakkeja näkyi
nyt melko tiheässä. Upseeri oli tsaarin vankilassa harvinainen.
Edelleen: yhtä liikuttava likaisuus kuin ennenkin ja aivan kysymättä
löysi "makuuhuoneen" vierestä saastaisen käymälän. Entiseen
tapaan tarjosi joku vakiintunut asukas teemukin ja utelias joukko
kertyi ympärille kuulemaan viimeisiä uutisia "vapaasta maailmasta".
Se "parempi huone", johon minut seuraavana aamuna siirrettiin,

poikkesi edellisestä vain siinä, että sen asukkaat — 135 miestä
tavallisen salin kokoisessa huoneessa — olivat sivistynyttä väkeä,
vastavallankumouksellisia (siis valtiollisia) ja keinottelijoita.
Todellakin: valtiollisia. Ennen oli valtiollisia ollut koko Venäjän
parhaimmisto. Niin oli nytkin. Suuri joukko samoja henkilöitä, jotka
ennen olivat tsaarihallituksen lähettäminä, viruneet vuosikausia
vankiloissa ja Siperian erämaissa. Nyt oli vain lisäksi tullut joukko
entisiä upseereita, lääkäreitä, professoreita, asianajajia y.m.
henkisen ammatin harjoittajia.
Vaikea on päätellä, kuinka vaarallisia nuo

vastavallankumoukselliset neuvostohallitukselle olivat. Vankilassa
puhutaan paljon ja näköjään hyvin avomielisesti, mutta vanhana
vankilan asukkaana olen saanut sen kokemuksen, ettei pidä liiaksi
luottaa kohtalotoverin puheisiin, semminkin jos hän on venäläinen.
Osa entisistä Siperian matkalla tapaamistani "anarkisteista" ja
muista "nisteista" paljastui ennenpitkää taskuvarkaiksi ja aivan hyvin
saattoi näilläkin "tovereilla" olla jotakin omallatunnollaan, mutta
miesten näöstä ja luonteenominaisuuksista päättäen lienee heidän
vastavallankumouksellisuutensa supistunut johonkin varomattomasti
lausuttuun löysään sanaan, tekoon heistä monikaan tuskin lie
uskaltanut ryhtyä. Heidän itsenäisyyden puutteensa valtiollisessa
ajattelussa ja mahdollisen rangaistuksen pelkonsa oli silmiinpistävä.
Venäläinen kansallistunto ei koskaan ole ollut kehuttava.
Bolshevikien vastustajissakin sen heikkous oli niin ilmeinen, että
oikein säälitti. Ei vapaudessa eikä vankilassa kukaan uskonut
Venäjän omin voimin selviytyvän bolshevikeista, vaan odottivat kaikki
kärsimättömyydellä jonkun ulkovallan esiintymistä. Tulkoonpa vaikka
ukko mustalainen ja orjuuttakoon Venäjän ikeensä alle, kunhan vaan
vapauttaa maan tästä painajaisesta. Sellainen virsi kuului kaikkialla
ja sen mukaisesti syntyi mitä hurjimpia huhuja milloin minkin
ulkovallan odotettavissa olevasta sekaantumisesta Venäjän asioihin.
Näitä huhuja lienevät alkuunpanneet Ententen asiamiehet
kiihoittaakseen herkkäuskoisia venäläisiä.
Niin, surkeita pelkureita istui tshekan säilytyshuoneissa

vastavallankumouksellisen nimellä. Ja laitoksen tutkijat osasivat
erinomaisen hyvin taidon tätä luontaista pelkoa lisätä. Kuulusteluissa
käytettiin revolveria tutkittavan ohimolla, välistä molemmillakin,
uhattiin kuolemanrangaistuksella, luvattiin nappia painamalla poistaa
väliseinät, jonka takana muka seisoivat valmiina kivääriniekat
napsauttaakseen tutkijan viittauksesta miehen hengiltä j.n.e. mitä
mielikuvituksellisimpiin temppuihin saakka. Kun sitten vielä tapahtui,
että vankeja siirrettäessä Krestyn vankilaan, jossa pitempiaikainen
tutkintovankeus tai määrätty istuminen suoritettiin, vartiosotilaat
ryhtyivät illanhämyssä hauskuuttamaan itseään kuljetettaviensa
ampumisella — viranomaisten tieten tai tietämättä, sitä en voi sanoa
—, ja kun sanomalehdet näistä tapahtumista aamuisin kertoivat
selittäen syyksi vankien karkausyrityksen, lannistui
vastavallankumouksen "sankarien" luonto kokonaan. Melkein sanoin
kuvaamaton oli se häiriö, joka syntyi, kun akkunamme alla
toimeenpantiin kolme teloitusta. Mitäpä, jos me tavalliset kuolevaiset
olisimme kauhistuneetkin tuollaista tapausta; mitäpä, jos me
olisimme inhosta värähtäneet nähdessämme bolshevikineitosen
rientävän verilammikoilta vaaleita kenkiään suojellen itse
tapahtumasta vähintäkään välittämättä, päivällistunniltaan kansliaan
kirjoituskoneensa ääreeni Mutta eiväthän vastavallankumouksen
miehet toki saaneet säikkyä verta!
Vai olivatko he todellakin vain herkkäkielisiä suunpieksäjiä,

salonkikumouksellisuuden ritareita? Luultavasti. Heidän vuoksensa
tuskin olisi kannattanut ruutia tuhlata. Mutta hallitus oli ehtinyt
maistaa verta ja siitä juopua. Se jännitti jousen mahdollisimman
kireälle ja sai toiseltakin puolelta hereille voimakkaampia aineksia.
Hallitus oli myöskin julistanut leppymättömän sodan keinottelua

vastaan. Se olikin, kuten ensimäisessä luvussa on kerrottu, paisunut
aivan hirvittävään laajuuteen. Taistelussa turvauduttiin edesmenneen
santarmiston elkeisiin. Tshekan urkkijoita vilisi kaikkialla ja entisiä
nimeltäkin tunnettuja santarmikätyreitä oli otettu laitoksen
palvelukseen. Keinottelijain keskuudessa oli saalistus helppoa ja
todistukset mukavasti saatavissa. Tsheka oli varustanut
asiamiehensä vangituilta takavarikoiduilla rahtikirjoilla, jopa
rahoillakin, joten he voivat esiintyä oikean tavaran myyjinä, tai
ulkomaista valuuttaa vahvasti omistavina rahanvaihtajina. Kun sitten
kaupat oli tehty, ilmestyivät toiset agentit ja näppäsivät kiinni niin
ostajat kuin myyjätkin. Tshekan omat miehet pääsivät kohta
lähtemään hienoissa ajopeleissään — muilta sellaiset jo siihen
aikaan oli takavarikoitu — uusille saalistusmatkoille tullakseen taas
joskus vastaisuudessa todistamaan päämiestensä käskystä.
Ankarillakaan toimenpiteillä ei keinottelua sentään koskaan saatu

loppumaan. Myöhemmin tämä käsite laajeni siihen määrin, että
kaikenlainen ostaminen ja myyminen oli rangaistuksen alaista, mutta
sitä harjoitettiin uhkauksilla höystetyistä määräyksistä huolimatta.
Siihen pakoitti hätä, joka aina lukee omat lakinsa.
Rikoksellisuuttakin vastaan taisteltiin, se on myönnettävä, joskin

aluksi huonolla tuloksella. Hallitus ei kyennyt saamaan
järjestyslaitosta lähimainkaan tyydyttävään kuntoon. Katuryöstöt
jatkuivat entiseen tapaan, asunnot eivät lainkaan olleet turvattuja.
Venäläinen on mestarillinen murtovaras. Hän avaa amerikkalaisen
lukon kuin leikillä ja, jos salpa on tiellä, sahaa hän
silmänräpäyksessä kädenmentävän aukon oveen, ja sekin este on
voitettu. Isäntäväen poissaollessa tyhjennettiin monen huoneen
asuntoja aivan koneellisen nopeasti ulkona odottaviin kuorma-
autoihin, jotka useimmiten olivat kotoisin jostain sotaväenosastosta
tai yleisestä laitoksesta.
Rikollisuuden laajaa levenemistä auttoi luonnollisesti vanha,

piintynyt venäläinen pahe: lahjustenotto. Bolshevikivirkamiehistäkin
oli suurin osa hyvin persoa pienille sivutuloille; Heidän katseensa oli
rahalla helppo kääntää sivulle siksi aikaa, kun uuden järjestelmän
arvovaltaa järkytettiin epäilyttävillä puuhilla. Joskus sattui jokunen
virkamies joutumaan kiikkiin ja saamaan rangaistuksen, jollei hän
puolestaan ehtinyt lahjoa toveriaan. Muistuupa tässä mieleen eräs
vallankumouksellisen tribunaalin näihin aikoihin antama tuomio,
jonka mukaan muuan 37,000 ruplaa lahjuksina ottanut komissaari
sai 37 vuotta pakkotyötä. Siis tuhat ja vuosi. Neuvostovaltakunnassa
ei ole rikoslakia, siellä tuomitaan "vallankumouksellisen
omantunnon" mukaan. Tällä kertaa oli vallankumouksellinen
omatunto mitannut oikeutta mainitulla tavalla. Turhaa on kuitenkaan
luulla, että lahjusten otto ankarimpienkaan tuomioitten (on langetettu
kuolemantuomioitakin) kautta olisi lakannut. Venäläinen herennee
ennen syömästä, kuin hylkimästä hyväntahtoisesti tarjottua
ylimääräistä korvausta "vaivoistaan".
Kuten edellä jo on sanottu, käyttivät tshekan tutkijat sangen

karkeita keinoja kuulusteluissaan. He eivät olleet juristeja,
useimmassa tapauksessa huonosti kirjoitustaitoisiakin, vieläpä
raakoja ja henkilökohtaisen kostonhimon sokaisemia. Muistanpa,
kuinka kerran eräs tehtailija oli juuri vapautunut tshekan
säilytyshuoneesta. Käytävässä oli häntä vastaan tullut entinen
työläisensä, nyttemmin tutkija, ja kysynyt minne matka. —
Vapauteen, oli tehtailija vastannut. Odottakaa, oli tähän tutkija
lausunut, — minulla on känsäiset kädet, mutta ei ole rahaa, teillä on
rahaa, mutta ei känsiä käsissä, saatte vielä hiukan aikaa miettiä tätä
asiaa. Ja tehtailija, jota vastaan ei ollut mitään muuta muistuttamista,
mietti samassa kopissa kuin ennenkin tutkijan teoriioja parisen
viikkoa, jonka jälkeen hänet laitoksen päällikön Uritskin käskystä
uudelleen vapautettiin, tällä kertaa peräyttämättömästi.
Tämä Uritski muodosti muuten tshekan henkilökunnassa

miellyttävän poikkeuksen. Ollen itse lainopillisesti sivistynyt,
suhtautui hän tutkittaviinsa, joskin ankarasti, samalla
oikeudenmukaisesti. Vangit suorastaan iloitsivat joutuessaan Uritskin
kuulusteltaviksi. He tiesivät, ettei heidän hermojaan kiduteta ja että
asia joutuu. Muuan vanhahko eversti, jolle Uritski jo oli lausunut
vapauttavan tuomion, oli tämän kysymykseen: mikä olette
mielipiteiltänne, vastannut suoraan:
— Monarkisti.
— Todellakin… ja te uskallatte sen sanoa!
— En voi salata vakaumustani.
— Menkää, olette vapaa.
— Sittenkin…
— Niin, minulla ei ole mitään tekemistä teidän mielipiteittenne

kanssa, mutta jos niitä yritätte toteuttaa teossa, saatte vastata.
Olisivatpa kaikki tshekan miehet olleet yhtä oikeamielisiä kuin

heidän päällikkönsä! Sen pahempi, ei niin ollut laita, eikä Uritski
ehtinyt joka paikassa olla mukana.
IV.
Hallitussuunnan ja yleisen tilanteen kehittyminen v. 1918.
Kuten aikaisemmin on viitattu, ei bolshevistisen hallitussuunnan

esiintyminen, varsinkaan taloudellisella alalla, ollut alussa lainkaan
varmaa. Tuntuvin isku vanhalle yhteiskuntajärjestykselle oli maan
yhteiskunnallistuttaminen. Tämä uudistus ei vaikuttanut juuri mitään
kaupunkilaisväestön asemaan. Ja, kuten sanottu, kaupunkilaiset,
vieläpä ainoastaan pääkaupunkilaiset, ovat maan asioita hoitaneet,
niin sodan kuin rauhan ja vallankumoustenkin aikana. Maaseudun
talonpoika, sikäli kuin hän vähitellen alkoi maareformia ymmärtää, oli
tyytyväinen, ja tilanomistajaluokka, joka oli saanut kovimman
kolauksen, on aina ollut hyvin vähälukuinen, eikä sen merkitys
yhteiskunnallisena tekijänä ole pitempään aikaan ollut
minkäänarvoinen. Bolshevismi oli yhdellä kynänvedolla pyrkinyt
hävittämään feodalismin maasta.
Sellaista toimenpidettä katseltiin monilla tahoilla suopein silmin —

toisilla taas suhtauduttiin siihen välinpitämättömästi.
Teollisuuteen ei bolshevismi aluksi lainkaan kajonnut. Tehtailijat

saivat toimia entiseen tapaan omissa laitoksissaan. He käyttivät
tehtaitaan edelleenkin omaan laskuunsa. Muutaman kuukauden
kuluttua kyllä ruvettiin perustamaan niin sanottuja tehdaskomiteoita.
Nämä olivat kokoonpannut työläisistä ja olemassaolonsa ensiaikoina
toimivat vain ammattikomiteoina, s.t.s. hoitelivat työläisten omia
taloudellisia asioita. Sitten niille siirtyi tuotannon kontrolli. Venäjällä
tämä kontrolli johti ikäviin seurauksiin: koko maan teollisuuden
pysähtymiseen. Bolshevikeilla tuskin lienee ollut sellaista aikomusta.
Tämän kirjoittajasta on aina tuntunut uskottavalta, ettei heillä ollut
selvää kommunistisen yhteiskunnan kaavaa valmiina
vallananastukseen ryhtyessään. Heidät nosti valtiollisen myrskyn
hyökyaalto pinnalle ja he heittäytyivät tuulen mukaan ilman
kompassia. Tuskinpa heidän parhaimmatkaan aivonsa v. 1917
lopulla osasivat ajatella, että kumouksen vyöry leviäisi muihinkin
maihin, kuten sittemmin, muutamien kuukausien kuluttua — osaksi
heidän omalla avustuksellaan — Suomeen ja myöhemmin Itävalta-
Unkariin ja Saksaan. Vasta seuraava syksy syöksi heidät vinhasti
eteenpäin kommunismin tiellä.
Niin, bolshevikit lienevät aluksi aikoneet antaa teollisuuden kulkea

yksityiskapitalismin lipun alla. Tehdaskomiteat kuitenkin jouduttivat
kehitystä toiseen suuntaan. Ne rupesivat vähitellen sekaantumaan
yhä enemmän koko tehtaan talouteen. Tehtailijoille itselleen ei jäänyt
minkäänlaista sananvaltaa. Opillinen ammattitaito sai väistyä
johdosta työläisten suosioon päässeitten nousukasluonteisten
tusinamestarien pöyhkeyden tieltä. Isännistö huomasi, että työtehon
suunnattomasti vähentyessä heidän vastuunsa, joka edelleen oli
voimassa, muodostuu tuhoisan raskaaksi. Raaka-aineitten hankinta
oli huonojen kulkuneuvojen ja valtiovallan palvelijain lyhytnäköisen
sekaantumisen vuoksi käynyt yhä hankalammaksi. Samoin pyrki
valtiomahti taistelussaan keinottelua vastaan pistämään liian usein ja
aiheettomastikin nokkansa tehdasteollisuuden alalle, epäillen
harjoitettavan milloin minkinlaista vilppiä, ja velvoittamaan
tehtaanomistajia määräämään hinnat alapuolelle kannattavaisuuden.
Tällaisissa oloissa ei teollisuutta tarvinnut kansallistuttaa. Se
kansallistui omasta alotteestaan siten että, kapitalistit yksi toisensa
perään jättivät tehdaslaitoksensa komiteain käsiin.
Elinkeinoelämän aloista pysyi kauppa kauimmin yksityisten

käsissä. Sota-aika oli kaikkialla, eikä suinkaan vähimmin Venäjällä,
järkyttänyt liikemoraalia. Houkutus syrjäyttää entiset terveet
liikeperiaatteet, jos sellaisia lie Venäjällä ennenkään sanottavasti
ollut, oli liian suuri ja koko liike-elämä oli muuttunut keinotteluksi.
Edellisten hallitusten aikana ei tähän seikkaan oltu kiinnitetty
huomiota, mutta kun bolshevikihallituksen jo edellisten vallanpitäjien
vahvasti "tullaamina" perimät varastotkin, kuten edellä on
huomautettu, alkoivat huveta olemattomiin, täytyi uuden
valtiomahdin pakostakin ryhtyä holtitonta menoa hillitsemään.
Kauppapuodit sensijaan olivat edelleen auki ja niissä löytyi tavaraa
minkälaista hyvänsä. Tuonti kylläkin pian lakkasi, mutta vanhoja
varastoja riitti v. 1918 lopulle, jolloin kauppa vasta kansallistutettiin.
Niinpä siis bolshevikien hallituskauden ensimäisenä vuonna

saattoi yksityinen kansalainen elää vielä verrattain säädyllisesti, kun
hänellä vaan oli rahaa. Enimmän sitä oli keinottelijoilla, vähemmän
palkallaan elävillä.
Pahimmin oli kärsinyt upseeristo. Heitä oli, niinkuin ensimäisessä

luvussa on kerrottu, joutilaina ja kerjäläisinä, sanomalehtien myyjinä,
kengänkiillottajina, ajureina, katu- ja kahvilalaulajina. Kyllä kai
heistäkin suuri osa olisi myöhemmin voinut sijoittua bolshevikien
palvelukseen, mutta he eivät halunneet — poikkeuksista puhumatta.
Säännöllistä armeijaa ei uudella valtiovallalla ollut. Vanha armeija oli
osaksi aseissa yhdessä työläisistä muodostettujen joukko-osastojen
kanssa punakaartina, kuten näitä bolshevikeille uskollisia
sotilasjoukkoja aluksi kutsuttiin, osa, monta vertaa suurempi
edellistä, oli heti Brest-Litowskin rauhan solmittua, jopa sitä
ennenkin, hajaantunut kotiseuduilleen. Vielä varsinaisen puna-
armeijan muodostuttuakaan eivät vanhat upseerit suostuneet siihen
liittymään. Vasta Puolan-Venäjän sodan aikana nähtiin
huomattavampi määrä heitä puna-armeijan riveissä. Jonkunverran
vanhoja upseereita on tietenkin harhaillut pitkin Venäjää tuolloin
tällöin syntyvissä valkoisissa joukoissa, mutta ovat hekin viime
vuosina armahdusten perusteella palailleet neuvostovallan alueille,
vieläpä ryhtyneet vanhaan ammattiinsakin.
Monista virastoista olivat myös aluksi useimmat toimitsijat

poistuneet saboteeraajain ja samalla nälkäisten joukkoon. Posti ja
sähkölennätinlaitos muodostivat sentään poikkeuksen. Niissä säilyi
melkein muutoksitta entinen virkakunta, vaikka molempien laitosten,
varsinkin postin, toiminta koko nykyisen valtakauden ajan on ollut
huonojen kulkuneuvojen vuoksi varsin avutonta, ajottain
pysähdyksissäkin.
Vapaan ammatin harjoittajat — asianajajat ja lääkärit — toimivat v.

1918 aikana entisellään, mutta tuomioistuimet oli lakkautettu. Nyt oli
olemassa vain vallankumoustribunaaleja, joissa oikeutta jakoi ennen
mainittu "vallankumouksellinen omatunto" milloin minkin huomatun
bolshevikin hahmossa.
Liikkeitten henkilökunta työskenteli, kuten ennenkin, samoin

pankkien ja tehdaslaitosten. Pankeista kyllä annettiin säästöjä ulos
vain määrän perästä — aluksi tuhat ruplaa kuukaudessa — ja
säiliölokeroihin ei saanut koskea, mutta työtä niissä riitti, kunnes ne
v. 1918 loppupuoliskolla kansallistutettiin ja yhdistettiin
valtionpankkiin. Niitten henkilökunta lienee sittenkin jäänyt melkein
ennalleen.
Toimettomia ja toimeentulottomia oli sittenkin Venäjällä puheena

olevana vuonna aika runsaasti. Entä minkälainen oli se toimeentulo,
jonka bolshevikihallitus palvelijoilleen antoi? Vatsa oli alunperin se
paikka, jossa kansalainen saattoi varmimmin tuntea, että
"köyhälistöhallitus" oli ohjaksissa. Kaikille palkannauttijoille, olivatpa
he sitten työläisiä tai virkamiehiä, maksoi neuvostohallitus aluksi 500
ruplaa kuukaudessa. Tämä ei riittänyt edes yhdelle hengelle
vaatimattomimpaankaan elämään. Tärkeät komissaarit, joihin
neuvostovalta aina on turvansa perustanut, saivat kyllä omissa
virkapaikoissaan ruuankin ilmaiseksi tai mitättömästä maksusta.
Mutta työläiset ja pienemmät virkamiehet näkivät nälkää. Kauaakaan
ei ehtinyt kulua, kun jo työläisten ihastus "köyhälistövaltaan" alkoi
hälvetä ja uusien vallanpitäjien "luokkaluonne" käydä ilmeiseksi.
Pitkälle ei tämä "kaikille sama palkka" järjestelmä vedellyt, syntyi
luokitus, joka jo vuoden lopulla käsitti kolmisenkymmentä ryhmää.
Työläisillekö nyt vihdoinkin palkkaa nostettiin? Ei, vaan taaskin
komissaareille ja suurille herroille, pienten jäädessä edelleen
kitumaan. Luontaisedutkin turvattiin vastuunalaisille toimitsijoille
entistä ehommat. Niistä myöhemmin.
Valtio kyllä lupasi leipää kansalaisilleen, mutta ei kyennyt

antamaan mitään. Yhä uhkaavampana lähestyi nälkä…
Maailmansodan syttyessä v. 1914 oli Siperian radan varsille jäänyt

kuljetusvalmista viljaa suunnattomat määrät. Venäjän rautatieverkko
ja kulkuneuvot eivät milloinkaan ole olleet kehuttavat, varsinkin
Siperian yksi ainoa rata on valmistumisestaan saakka potenut
liikarasitusta. Kun sodan alussa kaikki liikkuva kalusto otettiin
sotilaskuljetuksiin, jäi vilja monen muun tärkeän tarvikkeen ohessa
mätänemään. Tällaisia asioita ei ehditty ajatella. Venäläinen on aina
ollut suuri optimisti. Japanin sotaan lähtiessään hän kehui lakillaan
viskaavansa keltanaamat pakosalle ja jo ennakolta ikuistettiin
vihollisen musertava tappio sanoin, kuvin, yksinpä lasten
leikkikaluinkin. Suursodan alussa luvattiin olla kahden viikon kuluttua
Berlinissä, mitäpä siis sellaista pikamarssia varten leipäkasoja
kokoilla! Kun tämän kirjoittaja v. 1915 alussa Siperiassa vakuutti
venäläisille, että teillä on kahden vuoden kuluttua nälkä, leväyttivät
nämä tilastot auki ja sanoivat rintaansa röyhistäen: vot, katso, leipää
on, mehän olemme ruokkineet koko Europpaa, tottakai me nyt itse
murkinassa pysymme. Mutta samaiset karkoitetut saivat jo vuoden
perästä tuntea lähestyvän puutteen itsessään Siperiassa, Venäjän
vilja-aitassa.
Seuraavat sotavuodet lisäsivät vielä Siperian radanvarren

varastoja. Talvikelillä talonpojat vetivät sinne leipäviljaa, mutta sen
edemmäksi se ei kulkenut. Suuret kulutuskeskukset joutuivat
elämään kokonaan Keski-Venäjän ja Ukrainan viljasatojen varassa.
Jo tsaarinhallitus oli loppuaikoina pakoitettu turvautumaan
säännöstelyyn ja Kerenski tätä menoa jatkoi, säätäen lopulta
viljamonopoolinkin, joka tosin ei valtiolle suuria tuottanut.
Bolshevikit hiihtivät aluksi Kerenskin latua, mutta viljankoonnin

tulos niukkeni päivä päivältä. Keväällä ja kesällä pieneni virallinen
leipäannos jokseenkin näkymättömäksi. Jaettiin 1/8 naulaa leipää,
välistä sama määrä kauraa — jyvinä. Tähän oli tietysti luonnolliset
syynsä. Sota oli niellyt hirveän määrän parasta työvoimaa
maaseudulta. Suuret alat olivat jääneet viljelemättä ja Siperian vilja-
aarteita ei nytkään kulkulaitoksen täydellisen rappeutumisen vuoksi

Bioinformatics Tools For Pharmaceutical Drug Product Development 1St Edition Edition Vivek Chavda Full Chapter PDF

Uploaded by

Copyright:

Available Formats

You might also like

Bioinformatics Tools For Pharmaceutical Drug Product Development 1St Edition Edition Vivek Chavda Full Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bioinformatics Tools For Pharmaceutical Drug Product Development 1St Edition Edition Vivek Chavda Full Chapter PDF

Uploaded by

Copyright:

Available Formats

Bioinformatics Tools for

Pharmaceutical Drug Product

Tools For Chemical Product Design From Consumer

Chemoinformatics and bioinformatics in the

Discovery and Development of Antidiabetic Agents from

Computation in BioInformatics S. Balamurugan

Systems Architecture. Strategy and product development

Practical Application of Supercritical Fluid

Design of Hybrid Molecules for Drug Development Michael

Tools for Teaching 2nd Edition

Wiley Global Headquarters

Limit of Liability/Disclaimer of Warranty

Library of Congress Cataloging-in-Publication Data

Cover image: Pixabay.Com

Printed in the USA

Part I: Bioinformatics Tools 1

5 Mass Spectrometry, Protein Interaction and Amalgamation

Part II: Bioinformatics Tools for Pharmaceutical

8.3.7 Absorption, Distribution, Metabolism, Excretion

10.3.3 Predictive Maintenance (PdM) 210

12.3.1 Spectroscopic Techniques 260

Part III: Bioinformatics Tools for Healthcare

13.6.5 Increase in Unemployment 299

14.6.1 Deep Neural Networks 331

16.1.1.3 Mononuclear Cells (MNCs) 375

For a new drug to be developed and brought to market, approximately

methods, protein interactions, peptide-based drug design and omics tech-

*Corresponding author: vivek7chavda@gmail.com

Keywords: Bioinformatics, artificial intelligence, machine learning, AI/ML,

Bioinformatics allows rapid molecular modeling of biological processes

other domains to build cutting-edge computational and informational

1.2.1 Limitations of Bioinformatics

1.2.2 Artificial Intelligence (AI)

problem-solving, processing, and the ability to manipulate data or self-­

product development through risk management, shortened production

Advantages of AI include providing real-time data management, error

1.3 Machine Learning (ML)

prediction, anomaly detection, automation, and ranking of information

b) Claims Databases for finding patients:

- Quality Control - Prediction of 3D

Figure 1.1 Applications of bioinformatics, AI, and ML in the pharmacy sector.

d) Enhancing Commercial market survey:

1.4 Conclusion and Future Prospects

30. Hamet, P. and Tremblay, J., Artificial intelligence in medicine. Metabolism,

Spurthy College of Pharmacy, Marasur Gate, Bengaluru, Karnataka, India

Keywords: Artificial intelligence, drug development, drug discovery, lead

*Corresponding author: drrupeshgautam@gmail.com

by iterative simulations of different compounds’ interactions with tiny por-

2.2 Artificial Intelligence in Drug Discovery

*Distinction *Fixing *Data pre- *Training

Figure 2.1 The process of constructing an AI planner in general.

of initial data have a significant effect on the performance of an AI sys-

2.2.1 Training Dataset Used in Medicinal Chemistry

2.2.2 Availability and Quality of Initial Data

Yes Is the training set’s No

Yes Is the effectiveness of No Do some changes

Figure 2.2 General flowchart to describe model optimization techniques.

case is that such procedures are prohibitively expensive or impossible to

problem-solving, processing, and the ability to manipulate data or self-

Distinction Fixing Data pre- Training